Insights into FcγR involvement in pain-like behavior induced by an RA-derived anti-modified protein autoantibody

Joint pain is one of the most debilitating symptoms of rheumatoid arthritis (RA) and patients frequently rate improvements in pain management as their priority. RA is hallmarked by the presence of autoantibodies against several post-translationally modified proteins (AMPAs) such as citrullinated, carbamylated and acetylated proteins. It has been suggested that autoantibody-mediated processes represent distinct mechanisms contributing to pain in RA. In this study we investigated the pronociceptive properties of monoclonal AMPA 1325:01B09 (B09 mAb) derived from plasma cells of a RA patient. We found that B09 mAb induces pain-like behavior in mice that is not associated with any visual, histological or transcriptional signs of inflammation in the joints, and not alleviated by non-steroidal anti-inflammatory drugs (NSAIDs). Instead, we found that B09 mAb is retained in dorsal root ganglia (DRG) and alters the expression of several satellite glia cell (SGC), neuron and macrophage-related factors in DRGs. Using mice that lack activating FcγRs, we uncovered that FcγRs are critical for the development of B09-induced pain-like behavior, and partially drive the transcriptional changes in the DRGs. Finally, we observed that B09 mAb binds SGC in vitro and in combination with external stimuli like ATP enhances transcriptional changes and protein release of pronociceptive factors from SGCs. We propose that certain RA antibodies bind epitopes in the DRG, here on SGCs, form immune complexes and activate resident macrophages via FcγR cross-linking. Our work supports the growing notion that autoantibodies can alter nociceptor signaling via mechanisms that are at large independent of local inflammatory processes in the joint.


Introduction
Bidirectional communication between the immune and nociceptive systems has been identified as fundamental for the initiation and maintenance of chronic pain (Bas et al., 2012;Bersellini Farinotti et al., 2019;Goebel et al., 2021;Jurczak et al., 2021;Wigerblad et al., 2016). Notably, the effector functions of autoantibodies have been highlighted in the pathogenesis of several chronic pain conditions (Dawes and Vincent, 2016;Lacagnina et al., 2021). Historically, antibodies were thought to contribute to pain indirectly through antibody-mediated complement activation and/or via stimulation of proinflammatory cytokines release. However, not all chronic pain conditions are associated with tissue inflammation and in some autoimmune conditions, patients report chronic pain prior to an established inflammatory state. Recent preclinical works have identified several mechanisms by which autoantibodies contribute to mechanical hypersensitivity in rodents without overt inflammation. These include indirect mechanisms, such as binding to satellite glia cells (SGC) (Goebel et al., 2021) in dorsal root ganglia (DRG) and osteoclasts in the bone (Jurczak et al., 2021), both of which subsequently release pronociceptive factors, as well as direct mechanisms through activation of nociceptors via binding neuronal epitopes (Dawes et al., 2018;Klein et al., 2012) and activation of neuronally expressed Fcγ receptors (FcγR) by immune complex (IC) (Andoh and Kuraishi, 2004;Bersellini Farinotti et al., 2019;Jiang et al., 2017;Liang et al., 2019;Liu et al., 2020;Qu et al., 2012;Qu et al., 2011;Wang et al., 2019).
Rheumatoid arthritis (RA) is hallmarked by the presence of autoantibodies including rheumatoid factor (RF) and anti-citrullinated protein antibodies (ACPA). Autoantibodies can be detected many years before the diagnosis, and titers typically rise closer to the time of diagnosis (Ramos-Remus et al., 2015;Rantapää-Dahlqvist et al., 2003). It has been suggested that autoantibodies play an important role also in RA pain pathology (Catrina et al., 2017), in particular as arthralgia (joint pain) is frequently present for some time, ranging from years to months, before disease onset. This suggests that autoantibody-mediated processes could represent a distinct mechanism of pain in RA. ACPAs are present in approximately 50-70% of RA patients and have been reported to recognize citrullinated antigens such as vimentin, fibrinogen, type II collagen, α-enolase and filaggrin (Brink et al., 2016;Cook et al., 1996;Joshua et al., 2016;Lundberg et al., 2008;Mathsson et al., 2008). More recently, autoantibodies directed towards other post-translationally modified proteins have been discovered, for example anticarbamylated protein antibodies (Shi et al., 2011), and anti-acetylated protein antibodies (Juarez et al., 2016). Although autoreactivities to citrullination, carbamylation and acetylation were originally considered three independent classes of autoantibodies (Trouw et al., 2017), convincing evidence have shown they are concurrently present in some RA patients (Grönwall et al., 2021). Moreover, these antibodies have demonstrated cross-reactivity on both poly-and monoclonal level in a clonally dependent manner (Lloyd et al., 2019;Reed et al., 2016;Sahlström et al., 2020;Steen et al., 2019), and therefore may be regarded as one group of anti-modified protein antibodies (AMPAs) (Kissel et al., 2020;Sahlström et al., 2020).
Fcγ receptors are membrane-bound glycoproteins predominantly expressed in immune cells that bind the Fc part of IgG. In mice, FcγRIIb is considered inhibitory, while FcγRI, FcγRIII and FcγRIV are activating receptors (Nimmerjahn and Ravetch, 2008). FcγRI is a high affinity receptor and capable of binding e.g. murine IgG2a and human IgG1 and IgG3 in both monomeric and IC form (Bruhns et al., 2009;Dekkers et al., 2017;Overdijk et al., 2012). The other FcγRs are only activated by binding aggregated IgG ICs (Ravetch and Bolland, 2001). Recently, it was identified that FcγRI and FcγRIIb are expressed in primary sensory neurons in rat and mouse DRGs (Bersellini Farinotti et al., 2019;Qu et al., 2011) and that administration of ICs to dissociated DRG neurons result in increased neuronal excitability, calcium influx and release of substance P and CGRP (Andoh and Kuraishi, 2004;Jiang et al., 2017;Qu et al., 2011).
In this study we have investigated the pronociceptive properties of the monoclonal antibody (mAb) 1325:01B09, an AMPA which was generated from antibody-secreting cells isolated from synovial fluid of an RA patient and which specifically binds citrullinated, acetylated and carbamylated peptides (Sahlström et al., 2020;Steen et al., 2019). In vitro studies have demonstrated that 1325:01B09 binds to stressed or IL-8-exposed fibroblasts-like-synoviocytes (FLS) and stimulates their migration in a citrulline-dependent way (Sun et al., 2019). Interestingly, 1325:01B09 was also shown to react with nuclear targets in apoptotic cells and in both human and murine activated neutrophils (Lloyd et al., 2019). Previously, we have demonstrated that systemic injection of 1325:01B09 induced mechanical hypersensitivity in mice (Jurczak et al., 2021). However, that study was primarily focused on 1325:01B09 in combination with another RA-associated antibody. Here, we focus on 1325:01B09 alone and further investigate the mechanisms by which it induces pain-like behavior.

Animals
All performed procedures were approved by the local ethical committees (Stockholm Norra Djurförsöksetiska nämnd, Comité Régional d'éthique en matière d'expérimentation animale Auvergne and Animal committee of University Medical Center Utrecht) and done in accordance with European Communities Council Directive for the care of laboratory animals (86/609/EEC) as well as with the International Association for the Study of Pain (IASP) guidelines. BALB/c male and female mice were purchased from Janvier labs (Le Genest-Saint-Isle, France). BALB/c wilde-type (WT) and FcRγ chain − /− mice (lacking the activating receptors FcγRI, III, and IV; (Takai et al., 1994), originating from the same breeding line, were maintained as homozygous mice in parallel and bred at the Comparative Medicine Biomedicum, Karolinska Institutet. C57BL/6 male and female mice used for flow cytometry experiments were bred and maintained at the animal facility of University Medical Center Utrecht. All mice were housed in standard cages in a temperature-controlled room with a 12-hour light/dark cycle, with food and water ad libitum. The number of animals used for each experiment is indicated in the figure legends.

Monoclonal antibodies
As previously described, CCP2-positive 1325:01B09 (B09) and 1325:04C03 (C03) monoclonal ACPA were derived from single antibody-secreting plasma cells from a synovial fluid sample drawn from the knee joint of a female CCP-positive patient with RA (11 years disease duration) with clinical symptoms of active joint inflammation. Ig variable regions were cloned into vectors and monoclonal B09 and C03 recombinantly expressed as human IgG1 (Amara et al., 2019;Lloyd et al., 2018;Steen et al., 2019). The control mAb 1276:01G09 (Ctrl IgG), was generated from a single memory B cell and recombinantly expressed as human IgG1as previously described (Amara et al., 2019;Lloyd et al., 2018;Steen et al., 2019). Murine chimera antibodies (1325:01B09) were generated by replacing human gamma and lambda/kappa constant regions with the murine IgG2a constant regions. Murine IgG2a and human IgG1 bind to murine FcγRs with similar affinity (Dekkers et al., 2017). The 1325:01B09 antibody binds citrullinated, acetylated and carbamylated peptides and is considered an AMPA. The 1325:04C03 antibody primarily binds citrullinated peptides with some binding to carbamylated antigens (Grönwall et al., 2021;Sahlström et al., 2020). The control antibody, 1276:01G09 has no detectable reactivity. All expressed antibodies were extensively quality-controlled using ELISA, SDS-PAGE, size-exclusion chromatography and endotoxin testing (Amara et al., 2019). Each mAb was injected intravenously at a dose of 2 mg/mouse in PBS (volume max 150 μl/mouse).

Behavioral tests
Mechanical hypersensitivity was assessed with von Frey's filaments (Marstock OptiHair) using the up-down method, and the 50% probability of withdrawal threshold was calculated (Chaplan et al., 1994). Filaments were applied to the plantar surface of the hind paws. A positive response was characterized by a brisk withdrawal/licking of the hind paw. A cutoff of 4 g was applied to avoid tissue damage. The 50% withdrawal thresholds were evaluated before and at different timepoints of the model and expressed as force in grams. In addition, the withdrawal thresholds were used to calculate the hyperalgesic index, which for each mouse defines the magnitude of antibody-induced sensitization as the area between an extrapolated line at the level of the baseline and the time-response curve after antibody injection (area above the curve). These values are obtained using the trapezoidal method, which is achieved by dividing the total area into the sum of smaller trapezoids (one for each of two subsequent timepoints). Therefore, higher hyperalgesic indexes indicate greater hypersensitivity.
Sensitivity to cold stimuli was assessed in the same environment as mechanical hypersensitivity. After habituation, a drop of acetone was applied to the plantar surface of the hind paw and the duration of the nocifensive behaviors (shaking, licking, biting, and lifting the paw) were recorded. The test was repeated three times on each paw and the average was calculated.
Heat sensitivity was examined using a modified Hargreaves box (Dirig et al., 1997). A radiant heat stimulus was applied to the hind paw from below until a brisk withdrawal of the paw was detected by a motion sensor, stopping the stimulus and recording its duration. A cutoff of 20 s was applied to avoid tissue damage. The test was repeated three times on each paw, averaged and presented as latency (in seconds) to withdraw. For mechanical, cold and heat hypersensitivity animals were habituated twice to the test environment which was then followed by at least 3 baseline measurements performed on separate days. Animals were randomly assigned to control and antibody groups as well as treatment groups. The investigators performing behavioral tests were blinded to the experimental groups and treatments throughout the duration of the study.

Pharmacological experiments
Buprenorphine (0.1 mg/kg, Sigma-Aldrich) was administered by intraperitoneal (i.p.) injection on day 10 post-B09 injection and mechanical withdrawal thresholds were assessed at 2 and 6 h post injection. Following a two-day washout period, naproxen (50 mg/kg, Sigma-Aldrich) was repeatedly administered by subcutaneous (s.c.) injections on days 12, 13 and 14. Mechanical withdrawal thresholds were assessed each day before and 3 h after administration of naproxen. In a separate cohort of mice, 14 days post B09 injection, animals received a single i.p. injection of diclofenac (30 mg/kg, EliLilly). Mechanical withdrawal thresholds were assessed before, and 3 and 6 h after drug administration.

Joint histology
Hind ankle joints were post-fixed in 4% paraformaldehyde (PFA, Histolab) for 48 h. Tissue was then decalcified in 10% EDTA (Sigma) for 4-5 weeks, dehydrated in ethanol and embedded in paraffin. Sagittal sections (5 μm) were cut and stained with hematoxylin and eosin (H&E, Histolab). For each animal, 3 sections were analyzed by two blinded investigators for signs of synovitis, bone erosion and cartilage destruction using a 3-grade scoring system for each aspect, as previously described (Bas et al., 2012).

Immunohistochemistry
Fourteen days post-injection of either 2 mg of human B09 or sterile 0.01 M PBS animals were deeply anesthetized with isoflurane and transcardially perfused with PBS followed by 4% PFA.

Ankle joints
Ankle joints were harvested, post-fixed for 48 h in 4% PFA and decalcified using 10% EDTA solution at 4 • C for two weeks or until tissues were completely decalcified, characterized by plain X-ray (Fona X70, Fona). Decalcified joints were cryoprotected in 30% sucrose for 48 h and embedded in optimal cutting temperature (OCT) compound (Sakura Finetek). Longitudinal 20 μm sections were cut via cryostat (Leica 1900, Leica Biosystems) and mounted on gelatin-coated slides. Sections were then incubated with blocking solution consisting of 3% normal donkey serum (NDS), 0.3% Triton X-100 in 0.1 M PBS for 2 h and incubated with a primary antibody against CD68 (Rat anti-Mouse CD68, Biorad, MCA1957) in antibody solution (1% NDS, 0.1% Triton X-100) overnight. The following day, slides were washed three times with 0.1M PBS and incubated for 3 hours with anti-human Cy3 (Jackson Immu-noResearch, 1:300) and anti-rat Cy2 (Jackson ImmunoResearch, 1:300) in antibody solution. Slides were washed three times with PBS, counterstained with DAPI for 5 min, and then dehydrated in an alcohol gradient (70, 80, 90 and 100%, 2 min each), rinsed in xylene and cover slipped with DPX mounting medium. For each given marker, slides were scanned at low resolution to identify the ankle synovium. Confocal Zstack images were acquired at a 40× magnification with a Zeiss scanning confocal laser microscope (model LSM 800, Jena, Germany), raw stacks were processed using Zen 3.6 software by performing the maximum projection extended depth of focus function. IgG intensity was analyzed using the DRGquant pipeline to separate and organize image channels . IgG images were then analyzed in FIJI using a macro that analyzed all pixels above background for each image stack. The data was then concatenated into a single data frame using python and transferred to prism (9.4.1, Graphpad San Diego CA) for statistical comparison and graphing.

Image analysis of IgG accumulation in DRGs
Images were analyzed using the workflow outlined in DRGquant . In brief, the channels of raw.czi images were split using a python script. Images containing glutamine synthetase (GS) were then fed through a 2D Unet (Ronneberger et al., 2015) model trained to identify satellite glial cell signals. The GS images were run through a 2D Unet model trained to differentiate the neuronal soma rich region of the DRG from the fiber rich region of the DRG. Images were then run through a macro in FIJI (Schindelin et al., 2012) that identified satellite glial cells via connected components analysis using CLIJ (Haase et al., 2020) and returned quantifications of IgG intensities from within identified SGCs. Additionally, a summary image for each image stack was generated displaying a 2D projection of the raw images alongside SGC ROIs, which was visually inspected to assess and assure the quality of the analysis. Finally, a python script was used to concatenate all data tables generated as well as format outputs that could be used into GraphPad Prism (version 9.4.0, GraphPad Software).

Quantitative real-time PCR
Mice were decapitated under isoflurane anesthesia and ankle joints, together with the smaller joints (calcaneus, first and second metatarsal) of the hind legs and L3-L5 DRGs were collected and snap frozen at − 70 • C. For RNA extraction ankles were pulverized using BioPulveriser (BioSpec). Samples (joints and DRGs) were homogenized using Tissue-Lyser II (Qiagen) in TRIzol reagent (Invitrogen). Total RNA extraction was done according to the manufacturer's protocol and reverse transcription was performed using High-capacity cDNA Reverse Transcription Kit (Invitrogen). qPCR was performed using the pre-developed hydrolysis probes to measure the relative mRNA levels (Table 1). Data were normalized against housekeeping genes (HKG). Relative fold changes were calculated by the comparative Ct method (2-ΔΔCT).

Flow cytometry
Ten days post-B09 or saline injection, male and female mice were decapitated under volatile anesthesia and L3-L5 DRGs were collected for analysis of infiltrating immune cells. Tissue was gently homogenized and digested at 37 • C for 30-40 min with an enzyme cocktail (1 mg collagenase A with 0.5 mg trypsin in 1 ml DMEM, Sigma-Aldrich). Following pelleting, cells were incubated with anti-CD16/CD32 for 20 min at 4 • C, in order to block Fc receptors, preventing non-specific binding (Table 2). Next, cells were stained with various combinations of fluorochrome-labelled antibodies that target surface antigens, at 4 • C, for 20 min, in the dark. Cells were next fixed using fixation/permeabilization buffer for 15 min at 4 • C, in the dark, which was followed by incubation with antibodies targeting intracellular antigens for 20 min, at 4 • C, in the dark. Single color stains were done to determine the levels of compensation. Washes were done in-between each step using FACS buffer (PBS, 0.5% BSA and 0.1% NaN3 sodium azide). Counting beads were added to obtain the absolute number of cells, CD45 was used to gate out debris and dead cells. Samples were acquired by LSRFortessa flow cytometer (BD Biosciences) and analyzed with FACSDiva software. Forward scatter was used for all cellular analysis, as a trigger to identify events.

Cell culture 2.10.1. Macrophages
In order to generate primary macrophage culture, bone marrow cells were isolated from mouse tibia and femur. Cells were cultured in low adherence flasks in DMEM (Thermo Fisher Scientific) complemented with 10% FBS (Sigma), 2 mM GlutaMAX (Thermo Fisher Scientific), 50 U/mL penicillin, 50 μg/mL streptomycin (Invitrogen), and 10 ng/mL M− CSF (R&D) in a humidified incubator at 37 • C with 5% C0 2 . Upon reaching 70-85% confluency, cells were dissociated and seeded onto 8well Nunc TM Lab-Tek TM II CC2 chamber slide (Thermo Fisher Scientific) at a concentration of 1×10 5 /mL and placed in the incubator overnight.

Dorsal root ganglia (DRG) culture
Modified protocol based on Malin et al. (Malin et al., 2007) was used to generate primary neuronal and SGC-enriched/neuron-depleted cultures. In brief, C1-L6 mouse DRGs were dissected and kept in cold PBS until enzymatic digestion using papain (0.8 mg/mL; 30 min at 37 • C, Worthington Biochemical Corp.) followed by collagenase I (Worthington Biochemical Corp.) and dispase II (Sigma-Aldrich) (12 and 14 mg/mL respectively; 30 min at 37 • C). Cells were then gently triturated to form a single-cell suspension in Ham's F-12 Nutrient Mixture medium (Gibco) supplemented with 10% FBS (Gibco), 50 U/mL penicillin, 50 μg/mL streptomycin (Gibco). SGC-enriched/neuron-depleted cultures were obtained by seeding the cell suspension onto uncoated glass Nunc TM Lab-Tek TM chamber slides and removing the media with nonadherent cells (including neurons) 1.5 h post-seeding. DRG cultures for neuronal staining were seeded onto 8-well Nunc TM Lab-Tek TM II CC2 chamber slide (Thermo Fisher Scientific). Cells were maintained at 37 • C in 5% CO 2 atmosphere and allowed to recover for approximately 20 h before ICC. In order to stain the cells in non-permeabilized condition, cells were washed and stimulated with 100 μg/mL of human control IgG (control) or 100 μg/mL of human B09 IgG in media containing 10% FBS and 50 U/mL penicillin, 50 μg/mL streptomycin for 3 h in humidified incubator.

Culture stimulation and CXCL1 release
Satellite glia cell-enriched cultures were generated as described above and maintained by changing to fresh medium every two days. On day 5, cells were stimulated with 1325:01B09 hIgG1 (1 µg/ml) alone or together with different concentrations of LPS (5-50 ng/ml; Sigma-Aldrich). Supernatants were collected on day 6 (24 h post-stimulation) for subsequent mesoscale analysis. Levels of CXCL1 (KC-GRO, Meso Table 1 List of hydrolysis probes used for qPCR assay.

Statistical analysis
Differences between groups split into two independent variables were analyzed using two-way analysis of variance (ANOVA) followed by Bonferroni post-hoc test. Differences between two groups were analyzed using unpaired two-tailed student's t-test, while comparisons of three groups or more were assessed by one-way ANOVA followed by Bonferroni post-hoc test. All data are presented as mean ±standard error of mean (SEM) and p-value <0.05 was considered statistically significant. Statistical tests were performed using GraphPad Prism 8 (Graph-Pad Software).

Anti-modified protein antibody induces pain-like behavior in mice which is not attenuated by non-steroidal anti-inflammatory drugs (NSAIDs)
To examine the pronociceptive properties of B09 mAb, we injected 2 mg of human B09 or control antibody (i.v.) and assessed pain-like behaviors in female Balb/c mice. Mice injected with B09 mAb exhibited decreased mechanical thresholds, thus an increase in hyperalgesic index, compared to mice injected with control antibody (Fig. 1A). B09injected mice also displayed significant cold hypersensitivity on day 10 days, compared to mice injected with human control antibody (2 mg/mouse), which do not develop any hypersensitivity, n = 6/group. (B) Human B09 mAb induces cold allodynia on day 10 post-mAb injection, compared to control mAb, n = 8/group. (C) Human B09 does not induce significant heat hyperalgesia, measured on day 11 post-mAb injection and compared to control mAb, n = 8/group. (D) Murine B09 mAb (2 mg/mouse) induces significant mechanical hypersensitivity in mice, which lasts approximately three weeks, compared to saline-injected mice, n = 14-16/group. (E) Single injection of buprenorphine (0.1 mg/kg, i.p.) on day 10 post-B09 administration significantly increases mechanical thresholds at 2 h, but not 6 h post-drug injection, compared to baseline (BL) before drug administration, n = 4/group. (F) Repeated injections of naproxen (50 mg/kg, s.c.) on day 12, 13 and 14 post-B09 administration, do not reverse mechanical hypersensitivity in mice injected with B09 mAb, compared to BL, n = 4/group. (G) Single injection of diclofenac (30 mg/kg, i.p.) on day 14 post-B09 administration does not reverse mechanical hypersensitivity 3 or 6hours post-drug delivery, compared to BL, n = 6/ group. Data presented as mean ± SEM, * p < 0.05, **p < 0.01, ***p < 0.001 compared with control mAb or baseline (BL) before drug administration. (Fig. 1B) but no heat hypersensitivity on day 11 (Fig. 1C) post-B09 injection. To confirm that the hypersensitivity was induced also with B09 mAb with murine Fc, 2 mg of murine chimera B09 mAb was injected i.v. into female Balb/c mice. Similar to human B09, murine B09 mAb induced mechanical hypersensitivity that lasted approximately two weeks, as well as a higher hyperalgesic index, compared to salineinjected mice (Fig. 1D). To characterize the underlying mechanism of the antibody-induced hypersensitivity, we assessed the analgesic effects

Fig. 2. Systemic injection of B09 does not alter the expression of inflammatory factors in the ankle joint. (A)
Representative ankle joint sections stained with hematoxylin and eosin 14 days after B09 or saline i.v. injection. B09 does not cause significant synovial hyperplasia, cartilage destruction or bone erosion, as compared to saline control, n = 5-6/group. Expression of Tnf, Il1b, Il6, Il17Ra, Csf1, Itgam, Cxcl1, Il10, Ngf and Ngfr mRNA does not change in the ankle joint of male (B-L) or female (N-Y) mice, 15 days post-B09 injection, compared to saline or PBS control, n = 6/group. Expression of Cxcl2 was significantly upregulated in male joints (I), but not in female joints (V), compared to control. Ntrk1 was not detectable in either of joints (M,Z). Data presented as mean ± SEM, *p < 0.05, compared with control group. SAL, saline.
of buprenorphine, a semi-synthetic opioid, as well as NSAIDs naproxen and diclofenac. Only buprenorphine (0.1 mg/kg) administered 10 days post-B09 injection significantly alleviated mechanical hypersensitivity (Fig. 1E), while neither repeated injections of naproxen (50 mg/kg) on days 12, 13 and 14 post-B09 injection (Fig. 1F) nor a bolus injection of diclofenac (30 mg/kg) on day 14 post-B09 injection (Fig. 1G) had any significant analgesic effect. These findings suggest that the mechanisms of B09-induced mechanical hypersensitivity is independent of prostaglandin signaling.

Systemic injection of murine chimera B09 mAb does not alter the expression of inflammatory factors in the ankle joint
Although we did not observe any analgesic effect of NSAIDs on B09induced pain-like behavior, B09 mAb could still induce subclinical inflammation in the joints. Therefore, we examined ankle joints 14 days after i.v. injection of murine chimeric B09 mAb or PBS for histological or transcriptional indications of inflammation (Fig. 2). Histological analysis did not reveal any signs of synovial hyperplasia, cartilage destruction or bone erosion in joints from B09-injected mice, compared to PBS group ( Fig. 2A). We examined the effect of B09 on mRNA levels of cytokines, chemokines and other markers associated with inflammatory process, in the ankle joints of male ( Fig. 2B -M) and female mice ( Fig. 2N -Z). We did not detect any significant difference in the expression of Tnf, Il1b,Il6,Il17ra,Csf1,Itgam,Cxcl1,Il10,Ngf,Ngfr or Ntrk1 for either sex, and observed only a modest increase in Cxcl2 mRNA in male mice injected with B09, compared to control.

B09-induced mechanical hypersensitivity is dependent on Fcγ-chain receptors
To investigate if B09 mAb induced-pain-like behavior is dependent on activating Fcγ receptors, we injected FcγR chain − /− and WT mice with human or murine chimeric B09 mAb systemically. While both WT female ( Fig. 3A and B) and WT male mice (Fig. 3C) injected with 2 mg i. v. of murine chimeric ( Fig. 3A and C) or human (Fig. 3B) B09 mAb developed mechanical hypersensitivity, FcRγ chain − /− mice were protected from pain-like behavior ( Fig. 3A -C).

B09 mAb is not detected in the ankle joints but in the DRGs 14 days post-injection
In order to determine the main site of action of B09 mAb, we injected WT mice i.v. with 2 mg of human B09 or PBS and performed IHC in ankle joints 14 days post-antibody injection. Although we did not detect any human IgG in the ankle joints of B09-injected mice (Fig. 4A), similarly to PBS control, we found a significant retention of human IgG in the DRG (Fig. 4B). In order to compare the retention and binding of B09 mAb to another mAb, we used human C03 mAb, a different ACPA clone derived from the same RA patient. C03 mAb has a distinctly different reactivity pattern and impact on cellular function compared to B09, modulating e. g. osteoclast activity (Krishnamurthy et al., 2023;Steen et al., 2019). We detected both C03 and B09 mAb in the DRG 14days after injection, binding to structures that, upon qualitative inspection, resembled cells surrounding neuronal cell bodies, endothelial cells and fibroblasts (Fig. 4B). Further analysis revealed colocalization between B09 and glutamine synthase (GS, a marker for satellite glial cells, Fig. 4C, closeup) as well as much weaker colocalization with CD68, a marker for macrophages ( Supplementary Figure 1). In contrast, while C03 mAb had a much weaker colocalization with GS, it strongly colocalized with CD68 + macrophages in the DRG (Supplementary Figure 1). Assessment of signal intensity and cellular location using a modification of DRGqant revealed higher B09 mAb binding in the total DRG, based on the mean intensity of IgG binding (Fig. 4D). Importantly, signal in SGC soma was significantly higher for B09 compared to C03 mAb (Fig. 4D) but no difference in IgG intensity between B09 and C03 was detected in nuclei of either SGCs or other cells in the DRGs or the fibers. This suggests that both B09 and C03 mAbs gain access to the DRG in a naïve mouse, but only B09 mAb is significantly retained in the SGC soma in the DRG.

B09-induced upregulation of satellite glia cell and neuron-related factors in the DRGs does not depend on the presence of FcγRs
Because B09 mAb bound SGCs and accumulated in the DRG, we examined if systemic injection of B09 is associated with transcriptional changes in L3-L5 DRGs. Fifteen days post-B09 injection, we detected a significant increase in mRNA levels of Gs, S100b, Kcnj10 and Il17Ra (Fig. 5A -D), factors associated with SGCs (Ebbinghaus et al., 2017;Hanani, 2005;Miller et al., 2002;Vit et al., 2008). Interestingly, this increase was not dependent on the presence of activating FcγRs as the mRNA levels remained elevated in FcγR-deficient mice injected with B09 mAb (Fig. 5A -D). We next examined mRNA levels for neuronal factors and found that Ngf, Ntrk1, Bdnf, Tac1, Calca, Cacna2d1, Trpm8 and Trpv1 were elevated in DRGs from B09-injected WT mice compared to PBS (Fig. 5E -L). B09 injection caused a similar increase in mRNA levels of most of these factors in FcγR-deficient mice except for Tac1 (Fig. 5H) and Trpv1 (Fig. 5L), which were not different from PBSinjected WT mice.

B09-induced upregulation of macrophage and inflammation related genes in the DRGs is reversed by deletion of FcγRs
We have previously shown that while mouse DRG macrophages express FcγR1, this receptor is not present in SGCs or the cell surface membrane of neuronal cell bodies in the DRG (Bersellini Farinotti et al., 2019). This led us to examine if there were changes in macrophages and inflammation-associated factors. Intravenous injection of B09 induced a significant upregulation of several macrophage-related genes in the
DRGs: Itgam, Emr1 (F4/80), Aif1, Ctss and Cx3cr1 (Fig. 6A -E), while FcγR deficiency prevented increase in mRNA levels of these factors (Fig. 6A -E). We also found an increase in mRNA levels for factors typically released by macrophages such as Tnf, Il1b, and Cxcl1 as well as a tendency to an upregulation of Cxcl2 mRNA in the WT B09 mice, compared to control (Fig. 6F -J), while Il10 mRNA was not detectable in mouse DRG in either group. FcγR chain deficiency completely prevented the upregulation of Tnf and Il1b, but not Cxcl1 gene expression in the DRG, compared to control DRG. These findings suggest that some but not all B09-induced changes in the DRGs are dependent on the presence of FcγRs. Importantly, we did not find changes in inflammation or macrophage-associated factors in the lumbar spinal cord 14 days after injection of B09, compared to saline (Supplemetary Fig 2) indicating a local site of action of B09 in the DRG.

Human B09 mAb does not induce macrophage infiltration into DRGs and does not bind macrophages in vitro
Several animal models of chronic pain conditions are associated with an increased number of macrophages in the DRGs. Since we found a significant upregulation of macrophage-related genes in the DRGs, we were curious to understand whether B09 mAb causes immune cell infiltration into the DRGs. Thus, we performed flow cytometry on L3-L5 DRGs harvested 10 days post-B09 injection (Supplemetary Figure 3). We did not detect any changes in the number of immune cells (CD45 + , Fig. 7A), among which were monocytes (CD115 + , Fig. 7B showed that B09 mAb binds FLS, but only when subjected to inflammatory stimuli (TNF, IL-8) or serum starvation (Sun et al., 2019). Thus, we exposed macrophages to B09 mAb or control mAb (100 μg/ml final concentration) in media containing 10% FBS (normal condition, Fig. 7J) or 0.1% FBS (starved, Fig. 7K), to examine IgG binding to the cell surface prior to permeabilization during staining. While we saw some binding of control mAb in both non-starved and starved conditions, we could not detect any binding of B09 mAb to macrophages in either condition. Although these findings do not exclude that B09 induced functional changes in macrophages via FcγR activation, they did not reveal surface binding of B09 to macrophages.

Human B09 mAb binds satellite-glia cells in vitro
Since we found that B09 mAb colocalizes with SGCs in the DRGs, as well as enhances the expression of SGC-related factors, we next explored if B09 mAb binds SGCs in vitro. We dissociated DRGs and established SGC-enriched and neuron-enriched primary cultures which were incubated with human B09 or control mAb prior to cell fixation, to examine IgG binding in non-permeabilizing conditions. In SGC-enriched cultures, B09 mAb labelled some, but not all cells expressing GS (Fig. 8A). In contrast, no positive signal was detected in cells incubated with control mAb. In neuron-rich cell cultures, we could detect B09 binding to small, PGP 9.5-negative cells and no signal in either control mAb or negative control (Fig. 8B).
In order to understand if upon binding to the surface of SGCs, B09 mAb alters their activity, we designed an experiment where CXCL1 release into the cell supernatant was used as a readout of cell activity and LPS as a mean to "stress" the cells. Cells were incubated for 24 h with media containing 100 μg/ml B09 mAb and LPS (5-50 ng/ml).
While CXCL1 concentration in media of B09-stimulated SGCs did not differ from unstimulated cells (Fig. 8C), LPS increased the CXCL1 concentrations in media in a dose-response fashion. Furthermore, we found that co-incubation of B09 with 50 ng/ml LPS induced a significant increase in the CXCL1 concentration compared to LPS 50 ng/ml alone. In a separate experiment, we incubated cells for 16 h with antibodycontaining media (100 μg/ml of control mAb or B09 mAb), and then stimulated the cells with 50 μM ATP for 4 h. While antibodies alone did All genes but Tac1 and TRPV1 are upregulated in the FcγR-deficient DRGs, 15 days post-B09 injection, compared to control group. The gene expression data was normalized to Hprt mRNA levels and are presented as fold change (n = 4-7/group). Data presented as mean ± SEM, * p < 0.05, **p < 0.01, ***p < 0.001 compared with control group. not cause any changes in gene expression, the combination of B09 mAb + ATP significantly upregulated the expression of Cxcl1 (Fig. 8D), compared to ctrl mAb + ATP. Trends for increased expression of Cxcl2 (Fig. 8E) and Ngf (Fig. 8F) were also observed. These results suggest that B09 sensitizes to external stimuli like LPS or ATP and leads to enhanced transcriptional response or release of CXCL1.

Discussion
In the present study, we have replicated our previous findings showing that B09 mAb has a pronociceptive effect when injected into mice (Jurczak et al., 2021) and further elucidated the mechanisms through which this monoclonal AMPA induces pain-like behavior in mice. We have observed that B09-induced mechanical and cold hypersensitivity are not associated with any visual, histological, or transcriptional signs of inflammation in the joint and that mechanical hypersensitivity is not alleviated by single or repeated injections of NSAIDs. Instead, we found that B09 accumulates in the DRGs 14 days post injection and alters the expression of several SGC, nociceptor and macrophage-related factors in lumbar DRGs. Using transgenic mice that lack activating FcγRs, we uncovered that FcγRs are critical for the development of B09-induced pain-like behavior, and partially drive the transcriptional changes in the DRGs. Finally, we observed that B09 mAb binds SGCs in vitro and in combination with stimuli like LPS or ATP induces both transcriptional changes and protein release of factors linked to pain.
Previous pre-clinical studies from our labs have described the pronociceptive properties of several RA-associated autoantibodies, including polyclonal ACPA IgG isolated from RA patients (Wigerblad et al., 2016), anti-collagen and anti-cartilage oligomeric matrix protein antibodies (Bersellini Farinotti et al., 2019), monoclonal ACPAs in combination with LPS  and combination of two RA-associated autoantibodies, either B09 with B02 or B09 with C03 (Jurczak et al., 2021;Krishnamurthy et al., 2023). In the two latest studies, we found that B09 in combination with a second autoantibody induces bone loss (Jurczak et al., 2021;Krishnamurthy et al., 2023) and can drive pain-like behavior in mice via osteoclast (Jurczak et al., 2021) and peptidyl arginine deiminase 4 (Krishnamurthy et al., 2023) dependent mechanisms. However, B09 alone does not bind osteoclasts and has no impact on bone metabolism in vivo or in vitro (Jurczak et al., 2021;Steen et al., 2019;Sun et al., 2019). To confirm that the pain-like behavior observed with B09 was not dependent on a reaction towards human Fc or sex, we undertook studies with both human and murine chimeric B09 IgG and male and female mice and found that both forms of mAbs induce transient mechanical hypersensitivity with a similar time course in male and female mice. Similarly to what we have observed for other pronociceptive RA-associated mAbs at timepoints when they are not associated with overt joint inflammation (Bas et al., 2012;Christianson et al., 2010;Jurczak et al., 2021;Park et al., 2016), B09-induced pain-like behavior was not reversed by the NSAIDs naproxen and diclofenac, indicating a prostaglandin-independent mechanism. Instead, B09-induced hypersensitivity seems dependent on FcγR activation. Importantly, despite interspecies differences, IgG subclasses and FcγRs show substantial similarities and functional conservation between mammals. As such, human IgG subclasses are known to interact well with mouse FcγRs (Dekkers et al., 2017;Overdijk et al., 2012). Finally, while no change in gene expression of pro-nociceptive or proinflammatory factors was observed in the joints, except for a modest change in one out of eleven factors, prominent increases in mRNA for many factors were found in the DRG. This finding opens for several possible explanations for the pronociceptive properties of the B09 mAb.
It is noteworthy that FcγR-deficient mice are protected from B09induced pain-like behavior. Previous reports have shown that B09 mAb binds citrullinated antigens in plate-bound ICs that can readily activate FcγRs located on peripheral blood mononuclear cells (PBMCs) (Lloyd et al., 2019). Upon binding, B09-histone ICs stimulate PBMCs to The mRNA data was normalized to Hprt mRNA levels and were presented as fold change (n = 4-7/group). Data presented as mean ± SEM, * p < 0.05, **p < 0.01, ***p < 0.001 compared with control group. release pro-inflammatory factors like interleukin 8 (IL-8) and TNF. Furthermore, previous reports show that polyclonal RA-derived antibodies in plate-bound complex with citrullinated histone 2B or fibrinogen can stimulate cytokine release from macrophages in a FcγRdependent manner (Sohn et al., 2015;Sokolove et al., 2011). Although we cannot rule out that the pronociceptive actions of B09 mAb stem from IC-mediated activation of resident immune cells in the joint, the lack of detectable B09 in the joint 14 days post i.v. injection and lack of changes in mRNA levels of a broad panel of pro-inflammatory cytokines do not suggest such interaction after systemic delivery. Hence, if B09 has a local effect in the joint, a more likely scenario is that B09 reaches the joint and forms local ICs that induce mechanical hypersensitivity by activation of neuronally expressed FcγRs at an earlier timepoint than 14 days post i.v. injection (Bersellini Farinotti et al., 2019;Liu et al., 2020;Qu et al., 2011), and that B09 was cleared from the joint at the time of tissue collection. This would be in line with our earlier work demonstrating that anti-collagen type II antibodies in IC formation reduce mechanical withdrawal thresholds following ankle joint injection in an FcγRs-dependent and an inflammation-independent manner (Bersellini Farinotti et al., 2019) as well as in line with other reports of FcγRI expression on nociceptors and involvement in pain transduction (Andoh and Kuraishi, 2004;Bersellini Farinotti et al., 2019;Jiang et al., 2017;Liang et al., 2019;Liu et al., 2020;Qu et al., 2012;Qu et al., 2011;Wang et al., 2019).
Our observation that B09 mAb binds SGCs in vitro is intriguing especially in light of recent findings that pronociceptive IgG isolated from fibromyalgia patients can bind SGC in vivo and in vitro (Goebel et al., 2021;Krock et al., 2022). Due to their localization surrounding sensory neurons in the DRGs, SGCs are uniquely positioned to regulate neuronal excitability and emerging data suggest that they directly contribute to the development of various chronic pain states by allowing bidirectional communication and influencing nociceptive signaling (Hanani, 2012;Hanani, 2005;Hanani and Spray, 2020). Several studies have shown that SGC can undergo both functional and morphological changes in response to nerve damage or inflammation (Hanani and Spray, 2020). An increase in the proliferation and the number of gap junctions between SGCs, resulting in enhanced glial coupling that contributes to spread and enhanced neuronal excitability, have been observed in several models of neuropathic and inflammatory pain (Dublin and Hanani, 2007;Hanani et al., 2002;Nascimento et al., 2014;Ohara et al., 2008;Zhang et al., 2007). Activated SGCs are often associated with elevated Gfap, S100b, Il6 and P2x7r mRNA levels and release of molecules such as ATP, glutamate and cytokines, including CXCL1, which can act on neurons and increase their excitability (Afroz et al., conditions. Cells were live incubated for 1 h with either G09 mAb (100 μg/ml, control), B09 mAb (100 μg/ml) or only media (negative control), in order to examine only cell surface binding. Human IgG is labelled in red, Iba-1 in green, DAPI (nucleus) in blue. All scale bars indicate 25 μm. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 2019; Du et al., 2023;Dubový et al., 2010;Hanani and Spray, 2020;Huang et al., 2013;Souza et al., 2013;Takeda et al., 2007). While B09 mAb was previously reported to bind nuclear antigens under certain permeabilizing conditions, we have live-incubated DRG-dissociated cell cultures with the B09 mAb to investigate its cell surface binding. We have detected a positive binding signal in non-neuronal PGP 9.5-negative (presumably SGCs) in DRG cultures and GS-positive cells in SGCenriched cultures. Together with the fact that SGCs lack FcγRs (Qu et al., 2011), as well as the detected colocalization of B09 with SGCs in the DRGs 14 days post injection, suggests that B09 mAb could bind epitopes on the surface of these cells. This led us to explore if binding of B09 mAb to SGCs has any functional effects and contributes to pain-like behavior seen after B09 injection. We measured the release of CXCL1, a murine homologue of interleukin-8 (IL-8), which we considered as a marker of SGC activation in culture. While SGCs stimulated for 24 h with B09 mAb alone did not produce increased levels of CXCL1, a combination of B09 mAb and LPS augmented CXCL1 release into supernatant. It is possible that SGCs require an additional stimulus that renders them sensitive to B09 mAb. In order to confirm this hypothesis with a different stimulus, we have used ATP alone or in combination with B09 and/or control mAb. While we found that overnight incubation with B09 alone did not induce any transcriptional changes in the DRG cultures, a 4 h stimulation with ATP in the presence of the antibody, enhanced the expression of Cxcl1, Cxcl2 and Ngf. Importantly, this effect seems specific to B09 mAb as no significant changes were observed compared to the control mAb. It appears as if the presence of IgG in the culture dampened the stimulatory effect of ATP on the genes of interest, a mechanism that remains to be explained, but this effect was overcome by B09 compared to the control mAb. Intriguingly, a similar "secondhit" response pattern to B09 was seen with FLS, where antibody alone did not stimulate cell migration unless cells were exposed to starvation or certain proinflammatory signals (IL-8 or TNF) (Sun et al., 2019).
Systemic injection of B09 mAb triggered a global upregulation of several proinflammatory, glia-and neuronal-related factors in the lumbar DRGs. Remarkably, we could only detect upregulation of these genes in the sensory ganglia but not in the ankle joint or lumbar spinal cord. This demonstrates that inflammatory mediators such as antibodies can act locally in the DRGs and potentially promote pain-like behavior without inducing an overt systemic or local joint inflammation. It has been shown previously that the blood-DRG barrier has a much higher permeability to intravenously injected substances compared to the blood-nerve barrier and the blood-brain barrier (Hirakawa et al., 2004;Olsson, 1968). This unique feature of the DRG vascular system allows various endogenous and exogenous molecules like IgGs to enter the DRG parenchyma (Gunasekaran et al., 2018;Lund et al., 2023) and act directly on SGCs, resident macrophages and/or sensory neurons. It has been demonstrated that RA-associated IgG can be detected in the joint 24 h after i.v. injection in naïve mice (Jonsson et al., 1989), but we are not aware of studies comparing the vascular permeability between the joint and DRG. The targeted antigen(s) for B09 mAb in DRG remains to be identified, thus we can only assume that B09 binds to post-translationally modified antigens expressed on the surface of SGC cell membrane, causing a retention of B09 mAb in the DRG, compared to the joint. We did not observe hB09 binding to macrophages in vitro, but did observe hB09 colocalization with macrophages in vivo (Supplemental figure 1). Although we can't rule out that hB09 binds to epitopes on the surface of DRG macrophages in vivo, we recently found that a subset of DRG macrophages actively sequester macromolecules in circulation, including IgG (Lund et al., 2023). Therefore, our observations in vivo may well be the result of DRG macrophage phagocytosis of hB09. Furthermore, as the absence of activating FcγRs not only protected mice from pain-behavior, but also normalized the expression of most inflammatory and macrophage-related markers in the DRGs, it is plausible that the binding of hB09 to SGCs may serve as a substrate for IC formation and subsequent local activation of resident macrophages in the DRG via cross-linking of FcγRs on their surface. This, together with pronociceptive factors released from activated SGCs could result in enhanced neuronal excitability and hypersensitivity. The fact that FcγRs-deficiency did not alter the increased expression of satellite glia cell markers as well as some of the neuronal markers, like Calca or Trpm8, is interesting. It supports the notion that B09 activates SGCs in an antigen-dependent fashion, which is sufficient to alter gene expression in nociceptors. However, for a functional consequence, an IC-FcγR-dependent activation of DRG macrophage is required.
To summarize, our findings show that B09 mAb-induced pain-like behavior is dependent on FcγRs and point to that B09 mAb enters the DRG parenchyma and binds and activates SGCs. It is an intriguing possibility that B09 mAb forms ICs in the DRG, and thereby activates FcγRs expressed by resident DRG macrophages causing local release of cytokines ultimately enhancing sensitivity and excitability of peripheral .5 negative cells in DRG dissociated cell cultures. Cells were live incubated for 3 h with either control human mAb (100 μg/ ml), human B09 mAb (100 μg/ml) or only media (negative control), in order to examine only cell surface binding. Human IgG is labelled in red, PGP 9.5 or GS in green and DAPI (nucleus) in white. Scale bars in the first three upper rows of images ( Fig. 8A and B) indicate 20 μm, scale bars within the high-power images (fourth row) indicate 15 μm. (C) Analysis of the CXCL1 concentration in the supernatant of SGCs incubated with B09 and/or LPS, shows that B09 alone does not induce increased release of CXCL1 but in synergy with 50 ng LPS promotes significantly higher levels of CXCL1, compared to 50 ng LPS alone. Data presented as mean ± SEM, represents three independent experiments, *p < 0.05, compared with LPS 50. (D-F) DRG mixed cultures incubated with 100 μg/ml of control mAb or B09 and further stimulated with 50 μM ATP for 4 h, show a significantly enhanced expression of (D) Cxcl1, in response to B09 + ATP, compared to control mAb + ATP, as well as a trend towards enhanced expression of (E) Cxcl2 and (F) Ngf genes after B09 + ATP and compared to control mAb + ATP. Each point represents a technical replicate; data presented as mean ± SEM, * p < 0.05, **p < 0.01. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) nociceptors. While we have focused on one RA patient-derived AMPA in the current study, RA patients typically have a range of AMPAs with different fine specificities (Grönwall et al., 2021). Intriguingly, recent studies show a diversity of effects of such monoclonal antibodies on arthritis, with some of them, including B09 mAb, having an antiinflammatory and partially protective effect in the collagen antibodyinduced arthritis model (Raposo et al., 2023). This is in line with our earlier observation that the anti-collagen II antibody F4 induces mechanical hypersensitivity after systemic injection (Bersellini Farinotti et al., 2019), while the same antibody is protective when combined with other arithrogenic collagen type II antibodies (Croxford et al., 2010;Nandakumar et al., 2008). The present study on effects of the B09 mAb thus supports the notion that one and the same antibody can contribute differentially to pain and inflammation. Therefore, the role of AMPAs on different RA-associated symptoms, here on pain, are complex and can be postulated to differ depending on the autoantibody profile. Our work opens new avenues for the understanding of autoantibody-mediated pain mechanisms that are independent of joint inflammation and that may take place outside the joints, here in the DRG. Such mechanisms may contribute to the understanding of the common disconnect between arthritis and pain in RA, and thus to treatment of pain which cannot be explained by joint inflammation. Additional investigations are necessary to explore whether the inhibition of particular factors, which are elevated in the DRG, can attenuate pain-related behavior. More generally, studies of pain mechanisms caused by well-defined patient-derived antibodies will aid to disentangle the complexity of pain chronicity.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.