IgA Immune Complexes Induce Osteoclast-Mediated Bone Resorption

Objective Autoantibodies are detected in most patients with rheumatoid arthritis (RA) and can be of the IgM, IgG or IgA subclass. Correlations between IgA autoantibodies and more severe disease activity have been previously reported, but the functional role of IgA autoantibodies in the pathogenesis of RA is ill understood. In this study, we explored the effect of IgA immune complexes on osteoclast mediated bone resorption. Methods Anti-citrullinated peptide antibody (ACPA) and anti-carbamylated protein (anti-CarP) antibody levels of the IgA and IgG isotype and rheumatoid factor (RF) IgA were determined in synovial fluid (SF) of RA patients. Monocytes, neutrophils, and osteoclasts were stimulated with precipitated immune complexes from SF of RA patients or IgA- and IgG-coated beads. Activation was determined by neutrophil extracellular trap (NET) release, cytokine secretion, and bone resorption. Results NET formation by neutrophils was enhanced by SF immune complexes compared to immune complexes from healthy or RA serum. Monocytes stimulated with isolated SF immune complexes released IL-6 and IL-8, which correlated with the levels of ACPA IgA levels in SF. Osteoclasts cultured in the presence of supernatant of IgA-activated monocytes resorbed significantly more bone compared to osteoclasts that were cultured in supernatant of IgG-activated monocytes (p=0.0233). Osteoclasts expressed the Fc receptor for IgA (FcαRI; CD89) and Fc gamma receptors. IgA-activated osteoclasts however produced significantly increased levels of IL-6 (p<0.0001) and IL-8 (p=0.0007) compared to IgG-activated osteoclasts. Both IL-6 (p=0.03) and IL-8 (p=0.0054) significantly enhanced bone resorption by osteoclasts. Conclusion IgA autoantibodies induce release of IL-6 and IL-8 by immune cells as well as osteoclasts, which enhances bone resorption by osteoclasts. We anticipate that this will result in more severe disease activity in RA patients. Targeting IgA-FcαRI interactions therefore represents a promising novel therapeutic strategy for RA patients with IgA autoantibodies.


INTRODUCTION
Rheumatoid arthritis (RA) is a heterogeneous chronic inflammatory disease that is mainly characterized by swelling and pain in the joints. Approximately 0.5-1% of the population is affected and the incidence in women is higher compared to men. Distinct clinical phenotypes of RA can develop, which are thought to be a consequence of interactions between genetic and environmental factors (1). Autoantibodies often appear long before any signs of joint inflammation emerge (2)(3)(4)(5). The majority of RA patients develop antibodies against the Fc region of IgG (rheumatoid factor; RF). However, RF autoantibodies can also be found in patients with other rheumatic diseases and in non-rheumatic patients. Only IgM RF (and not IgG RF) is used in serological tests for RA. Antibodies against post-translationally modified proteins including anti-citrullinated protein antibodies (ACPAs) and anti-carbamylated protein (anti-CarP) antibodies are highly specific. ACPAs can be locally produced by plasma cells in joints of RA patients and are detected in sera of up to 70% of RA patients (6). They recognize citrullinated proteins or peptides, which can be found throughout the body including cartilage and bone. Anti-CarP antibodies are detected in sera of 45% of RA patients and recognize carbamylated proteins (7).
Currently, presence of ACPA and RF in serum are the most commonly used diagnostic markers for RA. Tested antibodies are predominantly of the IgG and IgM isotype. RF, ACPA and anti-CarP antibodies have been associated with bone loss in patients with established RA, suggesting a pathogenic role (3,7,8). Notably, bone loss has also been reported in ACPA-positive individuals who did not have joint inflammation, suggesting an adverse effect of ACPA on bone (9).
Bone remodeling is a process in which osteoblasts generate new bone and osteoclasts resorb bone to maintain bone homeostasis. This balance is lost in RA patients (10). Osteoclasts are large multinucleated cells, which form by fusion of mononuclear progenitors from the CD14 + monocyte/ macrophage lineage. Differentiation, survival and activity of osteoclasts depends on the presence of receptor activator of NF-kB ligand (RANK-L) and macrophage colony-stimulating factor (M-CSF) (11). Osteoclasts and their precursors express RANK, which interacts with RANK-L. The formation of osteoclasts is enhanced in an inflammatory environment by several cytokines, including interleukin (IL)-1b, IL-6, IL-8 and, in low doses by transforming growth factor (TGF)-b, which triggers RANK expression on osteoclast precursors (12,13). Increased levels of these cytokines have been found in synovial fluid of RA patients and have been reported to contribute to RA disease activity (14). These cytokines can be produced by inflammatory cells in the joint, such as monocytes and macrophages. Additionally, neutrophils are abundantly present and have been reported to express RANK-L. Their contribution in osteoclast activation is however debated (15,16). In the presence of autoantibodies neutrophils can form NETs and contribute to tissue damage by releasing their granule content (17).
Interestingly, neutrophils are particularly potently activated by IgA antibodies (18). Although they are not routinely tested in RA, more attention has been payed to IgA autoantibodies in RA in the last decade. A considerable proportion (~30%) of RA patients is negative for ACPA IgG and RF IgM (19). About one third of ACPA IgG and/or RF IgM seronegative patients had IgA autoantibody levels in their sera (RF, ACPA or anti-CarP) (3,7,19,20). Several studies have reported a correlation between the amount of IgA and disease severity in RA patients (21). Elevated levels of serum RF IgA were associated with extra-articular manifestations in RA patients and worse disease outcome (22,23). Moreover, patients having erosions in the bones of their hands had higher serum levels of RF IgA (24). Similarly, presence of anti-CarP IgA was associated with more severe disease (7). Nonetheless, the role of IgA autoantibodies on bone erosion is currently unknown. In this study we investigated the effect of IgA complexes and immune complexes isolated from synovial fluid of RA patients on bone resorption.

Patients
Synovial fluid from the knee (n=26) and serum (n=20) of RA patients was anonymously collected during their visit to the rheumatology department at the Leiden University Medical Center or the Amsterdam University Medical Center. Patients gave their written informed consent in accordance with the guidelines of the Medial Ethical Committee of the Leiden University Medical Center (reference no. B15.003) or the Amsterdam University Medical Center (reference no. 2013.234).

Precipitation of Immune Complexes from Synovial Fluid and Serum
Synovial fluids or sera were centrifuged at 3300xg for 5 minutes before use. Supernatants were mixed with an equal volume of 10% PEG6000 (Sigma Aldrich, St. Louis, MO) and incubated overnight at 4°C. Precipitates were collected after centrifugation for 10 minutes at 9500xg and resuspended in phosphate buffered saline (PBS; B.Braun, Melsungen, Germany) to the original volume of the synovial fluid/serum.

Serum Collection and Isolation of Human Monocytes and PMN
Serum was collected from healthy donors. Monocytes and polymorphonuclear leukocytes (PMNs) were isolated from peripheral blood obtained from healthy donors or from buffy coats (Sanquin, Amsterdam, The Netherlands). Monocytes were isolated using MACS CD14 MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany) according the manufacturer's protocol. In brief, peripheral blood mononuclear cells (PBMCs) and PMNs were collected after Lymphoprep (Axis-Shield, Oslo, Norway) density gradient centrifugation. PBMCs were washed in PBS and incubated with MACS CD14 MicroBeads after which CD14 + monocytes were isolated.

Monocyte Stimulation With Immune Complexes
CD14 + monocytes (2x10 6 /ml) were stimulated for 24 hours with BSA-, IgA-, or IgG coated beads in an effector target (ET) ratio of 1:100. Alternatively, 1x10 5 monocytes were stimulated for 24 hours with 10, 20 or 40 ml of immune complexes isolated from synovial fluid of RA patients. Supernatants were harvested and stored at -20°C.

NETs Release After Stimulation With Immune Complexes
PMNs (1x10 5 ) were incubated in RPMI with 1% FCS in black 96 well plates (FLUORTRAC 200, Greiner Bio-One) with 10 ml precipitated immune complexes from synovial fluid or serum from RA patients or healthy controls for 3 hours at 37°C. Release of extracellular DNA was detected by adding nucleic acid label SYTOX green (2,5 mg/ml; Invitrogen Life Technologies). Optical density was measured using a fluorimeter (FLUOstar/ POLARstar BMG Labtech GmbH, Offenburg, Germany) at 480nm excitation and 520nm emission.

Bone Resorption Assay
Bovine cortical bone was cut in slices of 0.5 mm thick and cut to fit in 96 well plates. Bone slices were sonicated, washed with PBS and incubated in a-MEM (Gibco BRL) for 30 minutes at 37°C after which 0.5x10 6 CD14 + monocytes resuspended in a-MEM supplemented with 10% FCS, 25 ng/ml M-CSF (ImmunoTools) and 50 ng/ml RANKL (BioLegend) were seeded in 96 well plates for 21 days. Supernatant of unstimulated monocytes or monocytes that had been stimulated with BSA-, IgA-, or IgGcoated beads for 24 hours was added (1:1) when indicated. Supernatants and medium were replaced twice weekly and when indicated, medium was supplemented with 25 ng/ml IL-6 (ImmunoTools), 10 ng/ml IL-8 or 5 ng/ml TNF-a (both Peprotech, Rocky Hill, NJ). After 21 days bone slices were stored in water (4°C) and bone resorption was visualized as previously described (26). In brief, bone slices were sonicated for 30 minutes in 10% NH 4 OH, washed in distilled water, shortly dried on filter paper and transferred to new wells containing water saturated aluminium potassium sulfate dodecahydrate (Merck Millipore, Burlington, MA) for 10 minutes. Bone slices were washed twice in distilled water, after which they were washed once under a strong current of distilled water and dried between filter paper. Bone slices were stained with Coomassie brilliant blue solution (PhastGel ® blue R, PlusOne Coomassie tablets, Pharmacia, Uppsala Sweden) for 1 minute, and dried between filter paper. Bone resorption pits were visualized with an inverted microscope (Nikon TE-300, Nikon, Tokyo, Japan) and quantified with Image J version 1.49v.

Purification and Stimulation of CD14 Negative Osteoclasts
Generated osteoclasts (7 day culture) were scraped from 48 well plates and incubated with MACS CD14 MicroBeads human (Miltenyi Biotec) according the manufacturer's instruction. The flow through containing CD14 negative cells was collected. To confirm purity of CD14 negative cells, cells were washed in 0.5% PBSA, blocked in 5% PBSA for 30 minutes on ice and stained with anti-human CD14-AF488 (Clone HCD14, Cat#325610, Biolegend) for 30 minutes on ice. Cells were measured with flow cytometry (LSRFortessa ™ X-20, BD Biosciences) and data was analyzed with FlowJo version 10.6.2 (Tree Star, Inc.). CD14 negative cells were resuspended in a-MEM supplemented with 10% FCS, 25 ng/ml M-CSF (ImmunoTools) and 50 ng/ml RANKL (BioLegend) and plated in 48 well plates for 24 hours. Medium was refreshed and cells were stimulated with 5 ml of BSA-, IgA-, or IgG-coated beads or 40 ml of immune complexes isolated from synovial fluid of RA patients for 24 hours, after which supernatants were collected and stored at -20°C. To determine the morphology of the sorted osteoclasts, TRAcP activity was measured and nuclei were stained with DAPI as described above.

Anti-Carbamylated Protein (anti-CarP) Antibody ELISA
Anti-Carp IgA and IgG were determined in synovial fluids of RA patients according to protocols previously described (7) with minor adaptations. Microtiter ELISA plates (ThermoFisher) were coated with 50 ml of 10 mg/ml carbamylated bovine serum albumin (Ca-BSA) or non-modified BSA and incubated overnight at 4°C after which wells were blocked with 1% PBSA for 6 hours at 4°C. Synovial fluid (1:10 diluted) was added for 1 hour overnight at 4°C. Wells were washed and incubated with F (ab')2 goat anti-human IgA-HRP (1:4000, ThermoFisher) or F (ab')2 goat anti-human IgG-HRP (1:4000, ThermoFisher) for 1 hour at 37°C. Fifty ml of TMB substrate was added after which the reaction was stopped with 50 ml of 10% H 2 SO 4 . Absorbance was measured with a microplate reader (Bio-Rad, Berkeley, CA) at 450nm.

Rheumatoid Factor (RF) Antibody ELISA
RF IgA levels were determined in synovial fluid of RA patients. Microtiter ELISA plates (ThermoFisher) were coated with 50 ml of 50 mg/ml human serum IgG (Sigma Aldrich) and incubated overnight at 4°C after which wells were blocked with 1% PBSA for 1h at 37°C. Synovial fluid (1:10 diluted) was added for 1 hour at 37°C. Wells were washed and incubated with F(ab')2 goat antihuman IgA-HRP (1:4000, ThermoFisher) for 1 hour at 37°C. Fifty ml of TMB substrate was added after which the reaction was stopped with 50 ml of 10% H 2 SO 4 . Absorbance was measured with a microplate reader (Bio-Rad, Berkeley, CA) at 450nm.

Statistical Analysis
Statistical analysis were performed with GraphPad Prism version 8.2.1 (GraphPad, San Diego, CA). Statistical differences were determined with unpaired Student two-tailed t tests (two groups) or one-way ANOVA (more than two groups) for normally distributed data. Mann-Whitney U tests (two groups) or Kruskall-Wallis tests (more than two groups) were applied to not normally distributed data. Correlations were determined with nonparametric Spearman correlation coefficients. P values <0.05 were considered statistically significant.

Autoantibodies in Synovial Fluid Induce Immune Cell Activation
To investigate whether local autoantibodies have a potential pathogenic role in RA, we first determined ACPA and anti-CarP IgA and IgG levels in synovial fluid of 26 RA patients ( Figure 1A). Since we previously demonstrated that RF IgA levels were significantly elevated in RA patients (27), we therefore additionally determined RF IgA in this study. Both IgA and IgG autoantibody levels showed great inter-patient variation. For every patient a profile of IgA and IgG autoantibodies present in their synovial fluid was created ( Figure 1B).
When neutrophils were stimulated with immune complexes isolated from synovial fluid of RA patients they formed NETs (Figure 2A). The level of NET formation varied greatly between samples. Immune complexes isolated from serum of either healthy controls or RA patients did not induced NET release (Figure 2A), which may have been due to lower concentrations of complexes in serum compared to SF. Of note, concentration of immune complexes in sera and SF of RA patients could not be determined due to limited patient material. NET release induced by SF immune complexes significantly positively correlated with ACPA IgA, ACPA IgG and anti-CarP IgG levels present in SF (Supplementary Figures 1A, B). Neither anti-CarP IgA nor RF IgA levels in SF correlated with NET release ( Supplementary  Figures 1B, C).
Next, monocytes were activated with immune complexes isolated from synovial fluid of RA patients and IL-6 production was determined ( Figure 2B). Monocytes were not stimulated with immune complexes isolated from serum of RA patients as neutrophils were not activated by serum immune complexes. Addition of a higher volume of SF immune complexes to monocytes resulted in increased IL-6 release. Furthermore, monocyte IL-8 production after stimulation with 40 ml of immune complexes was enhanced compared to unstimulated monocytes ( Figure 2C) and positively correlated with IL-6 release (p=0.0003) (Supplementary Figure 2A).
Immune complexes are large aggregates of autoantibodies and contain multiple antibody isotypes. ACPA IgA levels significantly correlated (p=0.0197) with monocyte IL-6 release induced by these immune complexes ( Figure 2D). Anti-CarP IgA and RF IgA followed a similar trend, but did not significantly correlate with IL-6 release (Supplementary Figure 2B). Neither ACPA nor anti-CarP IgG autoantibodies correlated with IL-6 release (Supplementary Figure 2C). Significant positive correlations between ACPA IgA (p=0.0011), ACPA IgG (p=0.0397) and RF IgA (p=0.0015) with monocyte IL-8 release induced by synovial fluid immune complexes were found ( Figure 2E). While anti-CarP IgA and anti-CarP IgG did not significantly correlate with monocyte IL-8 release (Supplementary Figure 2D).

Osteoclasts formation and activity
In vitro generated osteoclasts (OCs) are defined as large multinucleated cells, which express the enzyme tartrateresistant acid phosphatase (TRAcP) and can resorb bone (28). When osteoclasts mature, CD14 expression is lost and the expression of TRAcP increases (29,30). We generated multinucleated cells by culturing CD14 + monocytes for 7 days in the presence of M-CSF and RANK-L which expressed TRAcP ( Figure 3A). We defined osteoclasts as multinucleated CD14 negative cells with at least 3 nuclei and expressing high levels of TRAcP and aVb3. The aVb3 integrin enables the osteoclast to connect to the bone matrix (31). Not all multinucleated cells differentiated into osteoclasts as TRAcP and the aVb3 integrin were absent. These cells were likely multinucleated giant cells of macrophage origin. Since TGF-b is an important modulator of osteoclasts, we analyzed whether TGF-b influenced osteoclast formation. Addition of TGF-b to the 7 day culture resulted in formation of larger OCs with increased TRAcP expression ( Figure 3A). Moreover, presence of TGF-b resulted in the formation of more CD14cells (~80%) compared to absence of TGF-b (~20%) ( Figure 3B). Approximately 19% of the CD14population was classified as mature osteoclasts as they contained 3 or more nuclei ( Figure 3C). The formed OCs expressed high levels of the aVb3 integrin independently of TGF-b presence ( Figure 3D).
OCs cultured on bone slices for 21 days in presence of M-CSF and RANK-L resorbed bone as visualized by formed bone resorption pits and tracks ( Figure 3E, left panel). Presence of TGF-b significantly increased bone resorption by OCs ( Figure 3E, right panel). OCs in absence of TGF-b resorbed 5000 mm 2 bone, while OCs in presence of TGF-b resorbed significantly more bone (~15000 mm 2 ) (p=0.0445) ( Figure 3F).

IgA Activated Monocytes Induce Osteoclast-Mediated Bone Resorption
Supernatant of (activated) monocytes was added to osteoclasts which were plated on bone slices to study its effect on bone resorption. Unstimulated and BSA-activated monocytes supernatant served as control conditions and only induced little bone resorption by osteoclasts (respectively~3300 mm 2 and 7300 mm 2 ) as visualized by dark bone resorption pits ( Figure 4A). Supernatant of monocytes that had been activated with IgA complexes enhanced osteoclast induced bone resorption compared to supernatant of IgG-activated monocytes ( Figure 4A). OCs in presence of supernatant of IgA-activated monocytes resorbed on average~15000 mm 2 of the bone slices, while supernatant of IgG-activated monocytes only induced resorption of~2600 mm 2 bone by OCs (p=0.0233) ( Figure 4B).

Osteoclasts Express FcaRI and Are Activated by IgA
Next, we questioned whether presence of IgA autoantibodies also directly affects osteoclast activation. Human osteoclasts expressed intermediated levels CD16 (FcgRIII), CD64 (FcgRIII), CD89 (FcaRI), and low levels of CD32 (FcgRII) ( Figure 5A, upper panels). Subsequently, CD89 expression on multinuclear CD14 -avb3 + osteoclasts was confirmed with microscopy ( Figure 5B). Remaining CD14 + cells expressed high levels of CD16, CD32 and CD89, and intermediate levels of CD64 ( Figure 5A, lower panels). To investigate the effect of IgA-FcaRI interactions on the bone resorptive capacity of osteoclasts, we separated CD14 negative osteoclasts from the remaining CD14 positive cells. Of the sorted osteoclasts >80% was alive and >95% was CD14 negative (Supplementary Figure 3A). The majority of sorted OC were multinucleated and express TRAcP (Supplementary Figure 3B). OCs activated with immune complexes isolated from synovial fluid of RA patients produced IL-8 ( Figure 5C). However, similar levels of IL-8 were released by unstimulated osteoclasts. Production of IL-6 by osteoclasts after stimulation with SF immune complexes varied a lot between samples, but was not produced by unstimulated osteoclasts ( Figure 5D). To investigate whether cytokine production was mediated by IgA or IgG, we stimulated OCs with IgA-, or IgG-coated beads. IgAactivated OCs produced significantly higher levels of IL-8 compared to unstimulated, BSA or IgG-activated OCs ( Figure 5E). IL-6 was only produced by OCs after IgA activation ( Figure 5F). Bone resorptive capacity of osteoclasts was significantly enhanced in the presence of IL-8 (p=0.0054) or IL-6 (p=0.03) ( Figure 5G). These results support previous findings that IgA autoantibody levels in RA patients correlate with increased bone resorption. We now show that this can be a direct consequence of IgA-mediated activation of osteoclasts through the enhanced release of IL-8 and IL-6.

DISCUSSION
Bone remodeling in RA patients is disturbed, resulting in bone loss and eventually loss of function (10). In this study we demonstrated that IgA enhanced bone resorption through cytokines released by immune cells and by activating osteoclasts through FcaRI crosslinking. Our results show that IgA has a significant effect on bone resorption and in this way is likely responsible for worse pathology in RA patients.
The exact process of bone remodeling in RA patients is incompletely understood. It has been reported that multinucleated cells that express typical osteoclast lineage markers including TRAcP and cathepsin K are found at sites of bone erosion in RA patients (32). Multiple in vivo studies with osteoclast deficient mice support the critical role of osteoclasts in bone loss (33,34). Bone loss in RA patients was strongly associated with presence of ACPA or anti-CarP antibodies (7,(35)(36)(37). In contrast to our results, it was shown that bone resorption was enhanced by ACPA IgG containing immune complexes in vitro (38). Increased numbers of bone resorbing cells were responsible for this enhanced bone resorption and was also observed in presence of non-ACPA IgG containing immune complexes (38). Presence of ACPA IgA or IgG and anti-CarP antibodies in sera were predictive for the development of RA (3,7,39), whereas only ACPA-IgA levels were associated with more active disease (40). Next to osteoclasts, inflammatory cells present in the joint can indirectly contribute to bone resorption. Neutrophils are the most abundant immune cells present in the inflamed joint and can get highly activated by IgA and IgG resulting in the formation of NETs (18,41). NET release was induced by immune complexes isolated from synovial fluid of RA patients. Variations in the abundancy of NET formation is likely due to different levels of autoantibodies in SF, influencing the level of neutrophil activation. NET release was not induced by immune complexes derived from serum, which may be due to less presence of immune complexes compared to synovial fluid. NETs can contribute to the autoimmune profile of RA patients, as myeloperoxidase can convert thiocyanate into cyanate, which is essential for carbamylation (42). Previously it was shown that NET formation was significantly increased in neutrophils stimulated with synovial fluid containing ACPAs compared to stimulation with ACPA-negative synovial fluids (43). Similarly we reported positive correlations between NET release and ACPA IgA and IgG levels. NETs may further contribute to the pathogenesis of RA as source of citrullinated autoantigens. In complex with ACPAs they can enhance the production of pro-inflammatory cytokines IL-6 and IL-8 by local (immune) cells (41). Recently a humanized anti-ACPA therapeutic antibody was described to inhibit NET formation by human neutrophils activated with SF of gout patients (44), supporting the importance of this autoantibody in the pathology of RA as well.
Deposition of immune complexes is a common pathogenic feature of several autoimmune diseases, including RA (45). Immune complexes containing antibodies against altered proteins are suggested to initiate a state of non-resolving inflammation in the synovial cavity, followed by RF containing immune complexes reactive to these initial immune complexes (46). However, in animal models it has been shown that immune complexes do not have to be specific against altered proteins to be able to initiate synovitis (47). In the synovium immune complexes can augment inflammation through complement activation (48), Fc receptor engagement (49), toll-like receptor binding (50), and stromal cell activation (51). Furthermore, ACPA containing immune complexes are suggested to be recognized by membrane epitopes (e.g. citrullinated vimentin) expressed by osteoclasts which promotes osteoclastogenesis (38). The glycosylation status of autoantibodies present in immune complexes, such as ACPA sialylation, additionally enhances the capacity to promote inflammation (52). We showed that incubation of monocytes stimulated with SF immune complexes resulted in the production of IL-6 and IL-8. The variable levels of cytokine release can be explained by the different levels of autoantibodies present in the immune complexes. IL-8 is an important chemokine for neutrophils (53) and positively correlated with the amount of infiltrated neutrophils in the inflamed synovium of RA patients (43). IL-8 levels produced by monocytes correlated with both IgA and IgG autoantibodies. These results are supported by previous work from our lab demonstrating that IL-8 is produced by monocytes after IgA or IgG activation (own unpublished data). Interestingly, osteoclasts produced significantly elevated levels of IL-6 and IL-8 after stimulation with SF complexes. The direct effect of IgA autoantibodies on osteoclast activity by blocking of FcaRI could not be formally determined, as FcaRI is continuously internalized and re-expressed. Blocking FcaRI during long term culture of osteoclasts is therefore not feasible. Nonetheless, only stimulation of osteoclasts with IgA-, but not IgG complexes resulted in enhanced production of IL-6 and IL-8, supporting that IgA autoantibodies in SF were responsible for the observed IL-6 and IL-8 production. IL-8 after IgA, compared to IgG activation. Human CD14 + osteoclasts are described to express multiple Fc gamma receptors (54) and autoantibodies can bind osteoclasts specifically, thereby promoting osteolytic function in vitro (38). In this study we focused on CD14 negative osteoclasts, as the most mature bone resorbing cells (29), which have intermediate expression of FcgRI, FcgRIII, FcaRI, and low expression of FcgRII compared to CD14 + mononuclear cells. IgG autoantibodies were described to enhance the production of osteoclastogenic factors, including IL-6, IL-8, TNF-a and IL-1b by monocytes and macrophages (13,49). However, IgAactivation of monocytes induced higher levels of proinflammatory cytokines compared to IgG-activation (own unpublished data). Synovial fluid levels of IL-8 and IL-6 were significantly elevated in patients with high ACPA levels in their synovial fluid (43). ACPA-induced differentiation of osteoclasts was inhibited in vitro by a neutralizing IL-8 antibody. Furthermore, in vivo IL-8 inhibition reversed ACPA-induced bone loss in mice (55), supporting the importance of IL-8 in the pathology of RA in ACPA positive patients. IL-6 production by monocytes only positively correlated with levels of ACPA IgA, but not ACPA IgG. These results are in line with previous work showing IL-6 release by monocytes upon IgA, but not IgG activation (own unpublished data). Similarly, osteoclasts only produced IL-6 after IgA activation. IL-6 significantly increased bone resorption by osteoclasts. In vitro, IL-6 was shown to enhance osteoclastogenesis through activation of RANKL on osteocytes and osteoblasts (56,57). In vivo, IL-6 was necessary for inducing osteoclast activity and bone resorption (58,59). Several clinical trials have shown the potential of the human anti-IL-6 blocking antibody tocilizumab in treating RA patients (60)(61)(62).
In conclusion, we propose that IgA-FcaRI interactions on both infiltrated immune cells in the synovium and osteoclasts have an essential role in the pathogenesis of RA ( Figure 6). IL-6 production after IgA activation of monocytes and osteoclasts can contribute to osteoclast formation, activity and their bone resorptive capacity. Elevated IL-8 production after IgA stimulation of monocytes and osteoclasts can contribute to the infiltration of neutrophils, thereby promoting an inflammatory environment. Furthermore it promotes osteoclast activity resulting in increased bone resorption. We anticipate that targeting IgA-FcaRI interactions is a promising therapeutic strategy to diminish bone resorption in RA patients.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by The Medial Ethical Committee of the Leiden University Medical Center (reference no. B15.003) and the Medical Ethical Committee of the Amsterdam University Medical Center (reference no. 2013.234). The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
AB, MG, MD, and ME designed the experiments. AB, MG, and MD conducted the experiments. IJ contributed to the bone resorption assays. AB, MG, and MD analyzed the data. CL, TV, and ME provided scientific input. AB wrote the first draft of the manuscript. MG, MD, CL, TV, IJ, and ME revised the manuscript. ME supervised during the entire process. All authors contributed to the article and approved the submitted version.