Characterization of a synovial B cell-derived recombinant monoclonal antibody targeting stromal calreticulin in the rheumatoid joints

Abstract

Rheumatoid arthritis (RA) is characterized by formation of synovial-ectopic-lymphoid-structures (ELS) supporting B-cell autoreactivity towards locally generated citrullinated-(cit)-antigens, including those contained in neutrophils-extracellular-traps (NETs). However, only a minority of recombinant monoclonal antibodies (RA-rmAbs) from B cells isolated from ELS+RA tissues react against NETs. Thus, alternative cellular sources of other potential autoantigens targeted by locally differentiated B cells remain undefined. RA-fibroblast-like-synoviocytes (RA-FLS) have been implicated in the release of RA-associated autoantigens. Here, we aimed to define stromal-derived autoantigens from RA-FLS targeted by RA-rmAbs.

71 RA-rmAbs were screened towards RA-FLS by living-cell immunofluorescence (IF). Western blotting (WB) was used to identify potential autoantigens from RA-FLS protein extracts. Putative candidates were validated using co-localization IF-confocal microscopy, ELISA, immunoprecipitation assay and surface-plasmon-resonance on unmodified/cit-protein. Serum immunoreactivity was tested in ACPA+ vs ACPA- RA-patients.

10/71 RA-rmAbs showed clear reactivity towards RA-FLS in IF with no binding to NETs. One stromal reactive RA-rmAb (RA057/11.89.1) decorated a ~58KDa band which mass spectrometry and WB with a commercial antibody identified as calreticulin (CRT). Confocal microscopy demonstrated significant cellular co-localization between anti-CRT-RA057/11.89.1 in RA-FLS. RA057/11.89.1 was able to immunoprecipitate rCRT. Deimination of CRT to cit-CRT moderately increased RA057/11.89.1 immunoreactivity. cit-CRT displayed increased blocking capacity compared to unmodified-CRT in competitive binding assays. Finally, anti-cit-CRT antibodies were preferentially detected in ACPA+ vs ACPA- RA sera.

We identified a synovial B-cell derived RA-rmAb locally differentiated within the ELS+RA synovium reacting towards CRT, a putative novel autoantigen recently described in RA patients suggesting that FLS-derived CRT may contribute to fuel the local autoimmune response.

1. Introduction

Rheumatoid arthritis (RA) is the most common inflammatory erosive polyarthritis characterised by breach of self-tolerance and production of anti-citrullinated peptide/protein antibodies (ACPA). Highly mutated and Ig class-switched ACPA can be manufactured within synovial ectopic lymphoid structures (ELS) displaying features of functional germinal centres (GC) which develop in around 40% of RA patients (1–3). The frequent observation that hypermutated B cells within ELS in the RA synovium and other autoimmune conditions display evidence of clonal relationship and intra-tissue clonal diversification support the current notion that the humoral autoimmune response within ELS, such as those developing in the RA joints is antigen-driven leading to the local differentiation of autoreactive B cells (4–8).

Recently, we have shown that RA synovial recombinant monoclonal antibodies (RA-rmAbs) generated from single CD19+ synovial B cells isolated from ELS+ ACPA+ RA patients recognize locally released citrullinated antigens, such as those contained in neutrophil extracellular traps (NETs) (2). However, anti-NET immunoreactivity only accounts for a minority of the cellular immune-reactivity of the large amount of RA-rmAbs that we generated, leading to the hypothesis that alternative cellular sources exist which are responsible for the release of other potential autoantigens targeted by in situ differentiated B cells.

RA fibroblast-like synoviocytes (RA-FLS) play a crucial role in the pathogenesis of RA directly contributing to local cartilage destruction and synovial inflammation (9–14). RA-FLS are characterised by a sustained highly proliferative and activated state with increased level of anti-apoptotic and decreased level of pro-apoptotic factors which induce them to undergo hyperplasia (15, 16). Recently, RA-FLS have been shown to contribute to the local release of citrullinated antigens, particularly in the context of increased autophagy, suggesting that they may contribute to link local inflammation and autoimmunity by acting as an additional source of RA-associated autoantigens (17).

Thus, in this work, we aimed to investigate whether RA-rmAbs generated from single synovial B cells obtained from ELS+ ACPA+ RA patients display immunoreactivity towards RA-FLS and to identify putative stromal-derived autoantigens fuelling the local autoimmune response.

2. Patients and Methods

2.1. Patients

Synovial fluids and tissues from RA patients were obtained after informed consent (National Research Ethics Service Committee London - LREC 05/Q0703/198) by aspiration of swollen knees and from total joints replacement, respectively. RA patients were diagnosed according to the revised ACR criteria (Table 1) (18).

Table 1. Clinical data of the RA patients used for anti-cit-CRT antibodies ELISA assay.

The values are expressed as mean ± SEM. ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; VAS = visual analog scale of pain; DAS = disease activity score; ACPA = anti-citrullinated peptide/proteins antibody.

RA patients (n = 84)
Gender %	72.6 (F) / 27.4 (M)
Age	52.5 ± 1.7
ESR	38.6 ± 3.4
CRP	18.7 ± 3.5
VAS	64.9 ± 2.8
Tender Joints	10.3 ± 0.7
Swollen Joints	6.7 ± 0.6
DAS28	5.5 ± 0.2
CCP (or ACPA) antibodies	65 (ACPA+) / 19 (ACPA-)

2.2. Generation of RA-rmAbs from ELS+ RA synovial tissue

RA-rmAbs were generated from single synovial CD19+ B cells, as previously reported (2, 19). Compared to previous work, we obtained further 14 RA-rmAbs from one additional ELS+ ACPA+ RA donor bringing the total of RA-rmAbs tested to 80. Of these, we were able to express 71 RA-rmAbs at sufficient concentration for downstream analysis. Human monoclonal IgG from naïve B cells obtained from healthy donors (IgG-2c3) and rmAbs derived from naïve and memory B cells from Sjögren’s syndrome (SS) patients were used as controls (19).

2.3. Generation of fibroblast-like synoviocytes from RA patients and stimulation of NETosis

Fibroblast-like synoviocytes (FLS) were obtained either from synovial tissue or synovial fluid, as previously described (11, 20). At 90% confluent FLS were passaged 1:3 using 0.25% trypsin/EDTA (Sigma, UK). Culture medium was replaced every 3 to 4 days. FLS were used after passage 4 to avoid any contamination from synovial macrophages and up to passage 8. Neutrophils were isolated from peripheral blood (PB) of healthy donors using discontinuous gradient centrifugation and seeded onto cell culture cover slides at 2×10⁵ cells/well. Cells were activated with 100 nM Phorbol-Myristate-Acetate (PMA) for 4 h at 37°C to induce NETosis before fixation with 4% paraformaldehyde.

2.4. Immunofluorescence microscopy on FLS and neutrophil NET

FLS were seeded at 1x10⁴ cells/200 μl onto cover slides. After 24 h, cells were washed in 1X PBS and fixed using either ice-cold 1:1 acetone:methanol or 4% (final concentration) paraformaldehyde (PFA). After washing in Tris-Buffered Saline (TBS) and blocking with serum-free protein block (DAKO), RA-rmAbs or control rmAbs were diluted at 50 μg/ml in antibody diluent (DAKO) and applied for 1 h at RT. After washing with 1X TBS, Alexa-488 goat anti-human IgG was applied for 1 h RT. 4’, 6- DiAmidino-2-PhenylIndole (DAPI) (Invitrogen) was added to visualize the nuclei. All sections were visualized using an Olympus BX60 microscope. For double IF confocal microscopy in co-localization experiments, RA-rmAb or control rmAbs were incubated as above. A mouse anti-human calreticulin (CRT) (Abcam, clone: FMC 75; diluted 1:200) was then added for 1 h at RT. After washing, an Alexa-555-conjugated goat anti-mouse antibody (Invitrogen, 1:200) was incubated for 1 h at RT. After washing and mounting the slides were scanned using a Leica DM5500 confocal microscope. NET were stained with RA-rmAbs diluted in PBS for 1 h RT. After washing with TBS, Alexa488-goat-anti-human-IgG (Invitrogen, 1:200) was added for 30 min (RT). NETs were visualised by DAPI and cit-H4 using a polyclonal rabbit anti-histone H4 (citrulline 3; Millipore).

2.5. Protein extraction and western blot analysis

All procedures were performed at 4°C using pre-cooled reagents. FLS were washed in ice-cold 1X PBS. RIPA buffer (25 mM Tris-HCl pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) containing Protease Inhibitor Cocktail (Sigma) was added to the cell pellet. After 1 h on ice sample was centrifuged at 20,000xg for 10 min at 4°C to pellet the cell debris. Supernatant was collected and the protein concentration was measured using the BCA Protein Assay Kit following the Manufacturer’s instructions (Thermo Scientific).

500 ng of human recombinant calreticulin (hrCRT) (Abcam) or protein extract obtained from RA-FLS was loaded on 4-20% SDS-polyacrylamide gels (BioRad) and proteins were transferred to a nitrocellulose membrane (GE Healthcare Life Science). The blocking was performed in 5% (w/v) non-fat dry milk in 0.1% (v/v) Tween-20 TBS (blocking buffer) overnight at 4°C with gently agitation, followed by incubation with the primary antibody RA-rmAb or IgG-2c3 at 40 µg/ml in 5% blocking solution 2 h at RT with agitation. As a positive control for the Western blot, 40 ng of hrCRT was loaded on the gel and mouse anti-calreticulin antibody 1:2000 (Abcam) was used for the blotting. After rinsing 3 times for 10 min in TBS-T, the membranes were incubated with goat anti-human IgG peroxidase 1:10,000 (Jackson Immunoresearch) or goat anti-mouse IgG peroxidase 1:5,000 (Santa Cruz) in 5% blocking buffer for 1 h at RT. The membranes were washed again and incubated for 2 min in Clarity Western ECL substrate (BioRad). Band detection was performed using Hyperfilm ECL (GE Healthcare Life Science) and developed in a Konica Medical Film Processor (Konica Minolta). Densitometry analysis was performed using ImageJ software.

2.6. Immunoprecipitation experiment

Immunoprecipitation was performed mixing equal amounts (6 µg) of RA-rmAb or IgG-2c3 and hrCRT in 500 µl of Pierce IP Lysis Buffer (Thermo Scientific) on a rotary shaker for 2 h at 4°C. Protein A Sepharose beads (GE Healthcare Life Science) in IP Lysis Buffer were added to the mixture and incubated on a rotary shaker for 2 h at 4°C. After centrifugation and washing three times with cold IP Lysis Buffer, the immunoprecipitates were eluted with 1X Laemmli buffer and resolved using SDS-PAGE. Following the electrophoresis, the gel was stained using SimplyBlue SafeStain (Invitrogen) and the band around 58-kDa was excised and analysed by mass spectrometry, as described below.

2.7. Enzymatic digestion

In-gel reduction, alkylation and digestion with trypsin were performed on the excised gel bands prior to subsequent analysis by mass spectrometry. Cysteine residues were reduced with dithiothreitol and derivatised by treatment with iodoacetamide to form stable carbamidomethyl derivatives. Trypsin digestion was carried out overnight at RT after initial incubation for 2 h at 37 °C.

2.8. LC-MS/MS analysis

Peptides were extracted from the gel pieces by a series of acetonitrile and aqueous washes. The extract was pooled with the initial supernatant and lyophilised. The sample was then resuspended in 18μl of 50 mM ammonium bicarbonate to be analysed by LC-MS/MS. Chromatographic separation was performed using an EASY NanoLC system (ThermoFisherScientific, UK). Peptides were resolved by reversed phase chromatography on a 75 μm C18 column using a three-step linear gradient of acetonitrile in 0.1% formic acid. The gradient was delivered to elute the peptides at a flow rate of 300 nl/min over 60 min. The eluate was ionised by electrospray ionisation using an Orbitrap Velos Pro (ThermoFisherScientific, UK) operating under Xcalibur v2.2. The instrument was programmed to acquire in automated data-dependent switching mode, selecting precursor ions based on their intensity for sequencing by collision-induced fragmentation using a Top20 CID method. The MS/MS analyses were conducted using collision energy profiles that were chosen based on the mass-to-charge ratio (m/z) and the charge state of the peptide.

Raw mass spectrometry data were processed into peak list files using Proteome Discoverer (ThermoScientific; v1.4). Processed raw data was searched using the Mascot search algorithm (www.matrixscience.com) against the Uniprot database using All Taxonomy and Human Taxonomy.

2.9. Citrullination of calreticulin in vitro

hrCRT protein (Abcam) was incubated with rabbit skeletal muscle PAD (7.5U/mg) in 0.1 M Tris-HCl (pH 7.4), 10 mM CaCl₂, and 5 mM DTT for 2 h at 50°C. After incubation CRT was stored at -20°C. Citrullination was confirmed by Western blot analysis using an anti-citrulline (modified) detection kit (Merck Millipore) following the Manufacture’s instruction (Fig.S1).

2.10. ELISA assay for anti-CRT and inhibition assay

ELISA plates were coated overnight with unmodified/citrullinated CRT protein in 1X PBS at 1 μg/ml. RA-rmAbs or serum samples were transferred into ELISA plate and incubated for 1 h at RT. Unbound samples were removed before incubation for 1 h with horseradish peroxidase coupled goat anti-human IgG (1:5000). Assays were developed using TetraMethylBenzidine (TMB) Substrate Reagent Set (Becton Dickinson Optical Enzyme ImunoAssay (BDOptEIA)). Optical densities (ODs) were measured at 450 nm. All the RA-rmAbs and controls were tested at 50 μg/ml followed by 1:5 serial dilution (CRT protein only). Serum samples were tested after a 1:100 dilution. For the inhibition assay, RA057/11.89.1 antibody at different dilutions was pre-incubated with unmodified/citrullinated CRT protein (0.1 μg/ml) for 1 h at RT before being transferred to the unmodified/citrullinated CRT-coated plates. Thereafter, ELISA was carried out as described above. Results on RA sera were expressed as arbitrary units (AU). AU = (100/N) x OD_450nm serum sample where N is the lowest OD_450nm value in the anti-CRT antibody in the ACPA- RA patient group.

2.11. Surface Plasmon Resonance analysis via Biacore platform

All experiments were performed using a Biacore T200 instrument from GE Healthcare. Sensor chip Protein A, designated to bind human antibodies, and running buffer 10X HBS-EP+ were purchased from GE Healthcare. Running buffer was diluted 10 times with deionised water, filtered (0.22 μm) and degassed. RA-rmAb or control rmAb was immobilised on the sensor chip surface at 2 μg/ml for 30 s at a flow rate of 10 μl/min. CRT protein diluted at 500 nM, 250 nM and 125 nM in 1X running buffer was injected for 30 s at a flow rate of 10 μl/min. Running buffer was then flushed for 45 s at a flow rate of 10 μl/min, and finally the chip was regenerated by injecting a glycine solution (10mM, pH 1.5) for 30 s at a flow rate of 10 μl/min.

2.12. Statistical analysis

Differences in quantitative variables were analyzed by the unpaired (two sample) t Test and one-way ANOVA (multiple groups) using GraphPad-Prism5.01 software. Correlations were determined using “rcorr” function in R’s Hmisc package which computes a matrix of Pearson’s r and p correlation coefficients for all possible pairs of columns between two input matrices (i.e., anti-cit-CRT antibodies and each clinical data). Missing values were deleted in pairs. A p value <0.05 was considered statistically significant. Immunofluorescence colocalization analysis was performed by the Pearson Correlation Coefficient using ImageJ software (21).

3. Results

3.1. A subset of RA-rmAbs derived from synovial B cell clones target RA synovial fibroblast-like synoviocytes

71 RA-rmAbs generated from single synovial B cells (2) were tested for their reactivity towards RA synovial fibroblast-like synoviocytes (FLS) to assess whether the synovial B cell clones could target stromal-derived autoantigens. Immunofluorescence analysis showed that 10/71 (14%) rmAbs were uniquely reactive towards RA-FLS (Fig.1a-1b and 1c) suggesting that anti-FLS and anti-NETs antibodies are produced by largely independent populations of synovial B cells (Fig.1b). Conversely, none of the control rmAbs showed reactivity towards RA-FLS (Fig.1c). As depicted in representative images in Fig.1a, RA-rmAbs displayed a prevalent anti-cytoplasmic pattern in RA-FLS with invariably absent anti-nuclear immunoreactivity. As shown in Fig.S2, the RA-rmAb RA057/11.89.1 conserved the immunostaining on RA-FLS also in non-permeabilized RA-FLS suggesting that this antibody can also recognise the cell-surface form of CRT.

Figure 1. Synovial RA-rmAbs display immunoreactivity towards fibroblast-like synoviocytes (FLS).

(a) Representative immunofluorescence picture of RA-FLS and neutrophil-NETs incubated with different RA-rmAbs demonstrating selective immunoreactivity towards FLS-derived antigens (green). NETs were stained by DAPI (blue) and cit-H4 (red) using a polyclonal rabbit anti-histone H4 (citrulline 3; Millipore). (b) Representative immunofluorescence pictures of Human Umbilical Vein Endothelial Cells (HUVEC) cells incubated with the RA057/11.89.1 rmAb and the control rmAb IgG-2c3. (c) Pie chart summarising the RA-rmAbs reactivity towards RA-FLS (14%), FLS-neutrophil NETs (6%), NETs only (40%) and unknown reactivity (40%). (d) Representative immunofluorescence pictures of RA-FLS incubated with control rmAbs including IgG-2c3 and rmAbs from Sjögren’s syndrome patients.

3.2. Identification of calreticulin as an antigenic target of a specific RA-rmAb

In order to characterise the stromal autoantigens recognised by the RA-rmAbs, protein extract from RA-FLS was separated on SDS-PAGE, transferred on nitrocellulose membranes and probed with the RA-rmAbs. As shown in Fig.2a, one RA-rmAb (RA057/11.89.1) clearly displayed a strong reactivity towards a protein migrating in the ~58-KDa region. The band at 58-KDa was excised from the gel followed by trypsin digestion into peptides before mass spectrometry analysis (Fig.2b). LC-MS/MS analysis on the excised band corresponding to the 58-KDa molecular size confirmed the presence of human calreticulin (CRT) showing that CRT was the third most represented protein (from over 100 detected by the LC-MS/MS analysis) with a high amount of sequence coverage (62%) across the full length of CRT.

Figure 2. Calreticulin (CRT) expression in RA-FLS.

(a) RA-FLS protein extract was subjected to Western blotting and probed with anti-human-CRT monoclonal antibody or RA-rmAb. A protein of around 58-KDa is bound by the RA-rmAb RA057/11.89.1 and anti-CRT antibody. (b) Following mass spectrometry analysis by collision induced dissociation and database searching, peptide and protein assignment detected a high amount of sequence coverage (62%) across the full length of the CRT protein in the RA-FLS protein extract. 21 unique peptides were assigned from a total of 39 tandem mass spectra (MS/MS; highlighted in yellow). Modifications to particular amino acids are highlighted in green. (c) Representative CRT expression in RA-FLS from different donors (n=3) by Western blot. (d) Representative immunofluorescence pictures of RA-FLS showing expression of intracellular-CRT and cell surface cs-CRT (red). For intracellular-CRT, RA-FLS were fixed in ice-cold 1:1 acetone:methanol. For cs-CRT, RA-FLS were fixed in 4% paraformaldehyde (PFA). Nuclei were stained with DAPI (blue).

CRT is a conserved chaperone protein mostly expressed in the endoplasmic reticulum which migrates to a ~58-KDa position in SDS-PAGE (22). Alongside, we screened in silico for putative targets with similar expected molecular weight in SDS-PAGE based on the proteome analysis of RA-FLS performed by Dasuri et al (23). Interestingly, CRT emerged as one of the putative matches. Hence, a commercial mouse monoclonal anti-CRT antibody specifically recognised a band of overlapping molecular weight in RA-FLS protein extracts (Fig.2a).

We next confirmed the expression of CRT in RA-FLS from different donors by protein immunoassay and cell-based immunofluorescence. As shown in Fig.2c, CRT was found abundantly in RA-FLS protein extracts. We used human recombinant (hr)CRT to confirm the specific binding of the commercial anti-CRT antibody used in Western blot, although hrCRT displayed a slightly higher molecular weight compared to naturally occurring CRT in RA-FLS, probably due to post-translation modifications in E. Coli. Using immunofluorescence with a commercial anti-CRT antibody on living cells in permeabilizing and non-permeabilizing conditions, we demonstrated that CRT can be expressed by RA-FLS both intracellularly and on the cell-surface (Fig.2d).

3.3. The RA-rmAb RA057/11.89.1 targets FLS-derived calreticulin

We then confirmed that the RA-rmAb RA057/11.89.1 specifically target FLS-derived CRT. As shown in Fig.3a, double immunofluorescence staining with RA057/11.89.1 in combination with an anti-CRT antibody in permeabilizing conditions, analysed with confocal microscopy, demonstrated a strong cellular co-localization with CRT and the RA-rmAb which also recognizes CRT in RA-FLS protein extracts. The degree of colocalization between the two fluorophores was quantified using ImageJ by the Pearson correlation coefficient showing a strong correlation (r = 0.92). We next used hrCRT to screen the RA057/11.89.1 antibody by Western blot. As shown in Fig.3b, this RA antibody confirmed the binding towards hrCRT in Western blot whilst not only RA-rmAbs with no binding to RA-FLS in cell-based immune-screening but also other RA-FLS-reactive RA-rmAbs failed to recognise CRT in Western blot (Fig.3c and Fig.S3, respectively).

Figure 3. A specific RA-rmAb recognises unmodified CRT.

(a) Representative immunofluorescence picture showing staining for CRT (red) and RA-rmAb (green). Nuclei were stained with DAPI (blue). (b, c) RA-rmAbs binding to arg-CRT in Western blot. As negative control, a rmAb (IgG-2c3) from healthy donor naïve B cell was used. (d) RA057/11.89.1 RA-rmAb (red line) and negative RA-rmAbs binders (black lines) binding to arg-CRT by ELISA. All RA-rmAbs were tested at a concentration of 50 μg/ml followed by four serial dilution (1:5). Results are expressed as absorbance at 450 nm. The data are the results of two independent experiments. (e) Binding inhibition of RA057/11.89.1 RA-rmAb to arg-CRT pre-incubated with or without soluble arg-CRT (inhibitor). Results are expressed as percentage of binding inhibition. The data are the results of three independent experiments. **** p<0.0001

The binding towards unmodified hrCRT was quantitatively assessed by screening the RA057/11.89.1 in ELISA. The RA-rmAb showed binding to CRT in a dose-dependent manner (Fig.3d, red line). On the contrary, a large majority of RA-rmAbs failed to display any binding to CRT (Fig.3d, black lines). We used inhibition assay to further confirm whether CRT protein was recognized by the RA-rmAb. As shown in Fig.3e, pre-incubation of the RA057/11.89.1 antibody with CRT reduced the binding to unmodified CRT protein of around 60%.

In order to corroborate the CRT/RA-rmAb binding results, immunoprecipitation (IP) assays were performed. As shown in Fig.4a, IP of hrCRT with the RA-rmAb displayed a band around 58-KDa. Although we observed a similar band using the control rmAb IgG-2c3, because the heavy chain of the immunoglobulins (50-KDa) migrates in the same region of CRT (data not shown), LC-MS/MS analysis of the excised immunoprecipitate complexes clearly identified CRT in the IP CRT-RA057/11.89.1 sample but not in the IP CRT-IgG2c3 sample (Fig.4b-4c). Finally, binding to CRT was confirmed by surface plasmon resonance (SPR) via Biacore platform (Fig.4d).

Figure 4. Binding studies of RA057/11.89.1 towards CRT.

(a) Immunoprecipitation (IP) of human recombinant (hr)CRT and RA057/11.89.1 or IgG-2c3. As control, hrCRT alone was loaded. Coomassie staining is shown. (b, c) CRT protein was detected in the IP sample RA-rmAb/CRT following LC-MS/MS and database searching against the human portion of the Uniprot database but not in the IP sample with the control rmAb IgG-2c3. Example of a fragmentation spectra of a peptide from the CRT protein detected in the IP sample is shown. MS/MS fragmentation spectra of a peptide ion with mass-to-charge (m/z) ratio of 1043.99²⁺. Fragment ions annotated with the y-series are database assigned from the c-terminal end of the peptide while fragmentation peaks annotated with the b-series arise from the N-terminal end of the peptide. A strong consecutive matching of the peaks in both series provides strong evidence for the correct database assignment of the spectra to the peptide in the protein of interest. (d) Sensorgrams showing binding of RA057/11.89.1 RA-rmAb and one control rmAb to CRT protein used at different concentration. Binding is expressed as responsive unit (y axis) over time (x axis).

3.4. RA-rmAb binding characterization towards deiminated calreticulin

We next investigated whether the identified RA-rmAb with anti-CRT immunoreactivity displayed enhanced binding towards an in vitro citrullinated form of CRT (cit-CRT). CRT primary structure has 8 arginine residues which are potential sites of citrullination. Thus, unmodified CRT (arg-CRT) was deiminated in vitro by peptidyl arginine deiminase 2 (PAD2) and citrullination was confirmed by Western blotting using a specific anti-citrulline antibody (Fig.S1). We used both Western blot and ELISA to screen the RA-rmAb towards unmodified and deiminated CRT. As shown in Fig.5a, densitometry analysis of SDS-PAGE Western blot suggested that the anti-CRT RA-rmAb displayed an increased binding towards cit-CRT. Similar data were observed in ELISA towards citrullinated vs unmodified CRT, as depicted in Fig.5b. As represented in Fig.5c, pre-incubation of the RA-rmAb with deiminated CRT induced a higher decrease (around 60%) in the binding to in vitro deiminated CRT compared to pre-incubation with soluble unmodified CRT.

Figure 5. RA057/11.89.1 immunoreactivity towards deiminated calreticulin and expression of anti-cit-CRT antibodies in serum of RA patients.

(a) Left panel: RA057/11.89.1 RA-rmAb was tested in Western blot towards arg-CRT vs cit-CRT. Total arg/cit-CRT protein is shown in the top blot. Right panel: densitometry analysis of the Western blot is shown. Data were normalised towards total protein for arg-CRT and cit-CRT, respectively. (b) RA057/11.89.1 RA-rmAb binding to arg- and cit-CRT by ELISA. RA-rmAb was tested at a concentration of 50 μg/ml followed by four serial dilution (1:5). Results are expressed as absorbance at 450 nm. (c) Binding inhibition of RA057/11.89.1 RA-rmAb to cit-CRT pre-incubated with or without soluble arg- or cit-CRT (inhibitor). Results are expressed as percentage of binding inhibition. (d) Anti-arg-CRT and anti-cit-CRT antibody level in serum from ACPA+ RA patients (n=65), ACPA- RA patients (n=19), and healthy donors (HD; n=16) measured by ELISA. Results are expressed as arbitrary units (AU). AU = (100/N) x OD_450nm serum sample where N is the lowest OD_450nm value in the anti-arg-CRT antibody in ACPA- RA patient group. (e) Summary table showing correlation of serum anti-cit-CRT antibodies with ACPA, CRP, ESR, RF, VAS, tender/swollen joints and DAS28 score. ACPA = anti-citrullinated peptide/proteins antibody; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; RF = rheumatoid factor; VAS = visual analog scale of pain; DAS = disease activity score. The data in a, b and d are the results of two independent experiments while data in c are the results of three independent experiments. * p<0.05; ** p<0.01

3.5. Increased levels of anti-cit-CRT antibodies in sera from ACPA+ RA patients

We finally investigated the prevalence of anti-CRT antibodies in a cohort of 84 patients with early arthritis, naïve to any treatment, part of the Pathobiology of Early Arthritis Cohort (PEAC, http://www.peac-mrc.mds.qmul.ac.uk/) and 16 healthy donors (HD). Serum anti-cit-CRT antibodies were measured by ELISA using deiminated hrCRT (Fig.5d and Fig.S4). Early RA patients were divided in ACPA+ (n=65) and ACPA- (n=19) based on conventional anti-CCP2 test with healthy donors’ serum samples used as controls. As shown in Fig.5d, antibody levels to cit-CRT in ACPA+ RA patients were significantly increased compared to ACPA-. Analysis was performed to evaluate correlation between anti-cit-CRT antibodies in RA patient sera and the level of ACPA, CRP, ESR, RF, VAS, tender/swollen joints and DAS28 score at baseline (Fig.5e). A significant correlation was observed between anti-cit-CRT antibodies, ACPA (r = 0.26, p = 0.02), CRP (r = 0.27, p = 0.01) and ESR (r = 0.23, p = 0.04).

4. Discussion

The identification of ELS developing in the joints of RA patients as functional sites of B cell affinity maturation and the evidence that within ectopic GCs B cells undergo intra-synovial clonal diversification strongly indicated that humoral immune responses in the RA synovium are driven by locally released (auto)antigens (4–6, 8).

In order to investigate the cellular sources and the nature of the antigens recognized by hypermutated synovial B cells, we optimised a method to generate full rmAbs from B cells single sorted from ELS+ synovial tissues from ACPA+ RA patients. So far, we have generated over 80 RA-rmAbs which display for the vast majority highly mutated Ig H and L chain V genes and evidence of intra-tissue affinity maturation (2). In previous work, we identified a subset of around 40% of RA synovial B cells derived from ectopic-GCs which displayed reactivity towards antigens released by NETs, and characterised these autoantigens as primarily citrullinated histones H2A and H2B (2). In the present work, we explored the possibility that alternative, non-NETs, cellular sources exist in the RA joints capable of releasing other potential autoantigens driving the local adaptive immune response in the RA synovial tissue. In particular, we focused our attention on fibroblast-like synoviocytes (FLS), a key pro-inflammatory component of the RA synovitis, which contain high amount of putative RA-associated autoantigens in their deiminated form, such as vimentin and alpha-enolase, as shown in a proteomic profiling of RA-FLS (23). Additionally, recent work which we contributed to demonstrated that the induction of autophagy in RA-FLS favours the generation of citrullinated antigens, suggesting that RA-FLS may contribute to inflammation and autoimmunity also by releasing RA-associated autoantigens in the synovial microenvironment (17).

Therefore, we initially screened our RA-rmAbs using indirect immunofluorescence with live primary RA-FLS from different donors as substrate. By this mean, we identified 10 (14%) RA-rmAbs with clear immunoreactivity to RA-FLS without any binding to NETs. Of relevance, all the anti-FLS clones displayed a prevalent cytoplasmic pattern in immunofluorescent staining and for the vast majority were immunoreactive using both permeabilizing and non-permeabilizing methods.

Using immunoblot from RA-FLS protein extracts, we observed that one RA monoclonal antibody (RA057/11.89.1) was strongly reactive towards a ~58-KDa band. Analysis of the V(D)J gene usage for both H and L chains revealed that RA057/11.89.1 monoclonal antibody was characterised by V_H1-18/D2-2/J_H6 and Vκ3-20/Jκ4 gene segments. Furthermore, the original isotype of this clone was Igμ which sustained the low number of mutations observed in the HC variable region (n=1). Instead, we observed a higher number of mutations in the LC variable region (n=12). A detailed analysis of this clone is reported in Table 2.

Table 2. V(D)J gene usage and somatic mutations analysis of RA057/11.89.1.

FR = framework region; CDR = complementarity determining region; (-) / (+) = negative / positive charges.

RA057/11.89.1 IgM
Heavy Chain	VH	D	JH	(-)	CDR3 (aa)	(+)	Length
	1-18	2-2	6	1	RYCSSTSCYKGSYYYYYYYMDV	2	22
Light Chain	Vκ		Jκ	(-)	CDR3	(+)	Length
	3-20		4	0	QQYGSSPLT	0	9
Mutations	V-REGION Nb of mutations	FR1 Nb of mutations	CDR1 Nb of mutations	FR2 Nb of mutations	CDR2 Nb of mutations	FR3 Nb of mutations	CDR3 Nb of mutations
Heavy Chain	1	1	0	0	0	0	0
Light Chain	12	12	0	0	0	0	0

Mass spectrometry of RA-FLS protein extract, in silico analysis of RA-FLS proteomic profiles and a series of co-immunoblot and co-localization confocal microscopy experiments, identified the 58-KDa band as calreticulin (CRT). CRT is a conserved chaperone protein which migrates into the 58-KDa position in SDS-PAGE (22), mainly expressed in the endoplasmic reticulum and responsible for Ca²⁺ transportation and folding of glycoproteins (24). CRT can also be expressed on the cell surface playing a critical role in the clearance of apoptotic cells (25) and be released in the extracellular environment via the secretory pathway (26). It is formed by three domains: i) N-terminal domain, ii) middle domain named P-domain and iii) C-terminal domain. CRT has been found to be abundantly expressed in RA-FLS (23) and several studies indicated a higher concentration of CRT in the serum and synovial fluid of RA patients compared to osteoarthritis (OA) and HD serum samples which correlated with RA disease activity (27, 28). Increased levels of CRT in the synovial tissue of RA compared to OA patients have been also demonstrated (24, 25, 27–29). Interestingly, it has been shown that CRT recognizes the RA “shared epitope” (SE) HLA domain sequence and can modulate the signalling activated by the SE ligand when present in its citrullinated form (29). Moreover, although native CRT has been described as an autoantigen in several autoimmune conditions (30, 31), its role as a target of autoreactive B cells in RA has only very recently been investigated with the demonstration that around 60% of RA patients display circulating anti-cit-CRT antibodies (32).

In our work, we first confirmed that CRT was highly expressed in RA-FLS not only intracellularly but also on the cell-surface using a highly monoclonal anti-CRT antibody. The specific reactivity of one of our RA-rmAbs (RA057/11.89.1) with CRT was then confirmed by using at least 3 methods: i) co-localization with anti-CRT in confocal microscopy; ii) Western blot using RA-FLS protein extracts and/or hrCRT as substrates and immunoprecipitation using CRT and the RA-rmAb followed by LC-MS/MS analysis and iii) ELISA using hrCRT with competitive binding assays. We also generated cit-CRT by deiminating CRT with PAD2 in vitro and demonstrated using both immunoblot and ELISA that the anti-CRT rmAb identified displayed enhanced binding to the citrullinated compared to the native form of CRT, with pre-incubation with cit-CRT able to decrease RA057/11.89.1 immunoreactivity by 60%.

To confirm the results obtained at single synovial B cell clonal level with the systemic autoantibody production in RA patients, we tested the reactivity of 84 patients with early RA towards cit-CRT. We significantly detected anti-cit-CRT antibodies more frequently in the serum of ACPA+ RA compared to those with a negative ACPA status, expanding on recent data obtained in established RA patients (32), and suggesting that CRT acts as an autoantigen already in early stages of RA in a subset of patients. Interestingly, anti-cit-CRT antibodies were significantly and positively correlated with both ACPA, CRP and ESR levels, although their clinical significance in the context of the ACPA family remains to be elucidated in larger prospective cohorts.

Nevertheless, our work highlights CRT as a novel autoantigen locally released in the RA synovial compartment which appears to promote local humoral autoimmunity. These data are also of interest in line with recent studies showing that cell-surface CRT in its citrullinated form can enhance the binding to a SE ligand, and can activate downstream innate and adaptive immune cell signalling, which is in keeping with the notion that ACPA are strongly associated with the shared epitope amino acid sequence of the HLA-DRβ chain (SE) (25, 29, 33–35). Whether anti-CRT antibodies could interfere with the strength of signalling from the proposed CRT-SE complex remains to be formally elucidated, but it has been proposed that these antibodies might affect the binding of cit-CRT to the SE ligand, thus influencing the inflammatory cascade activated by this interaction (32). Likewise, further experiments are needed in order to investigate whether anti-CRT antibodies can modulate RA-FLS function and promote a pro-inflammatory phenotype in these cells. The characterization of RA057/11.89.1 as an anti-CRT rmAb provided in our work will pave the way for such functional experiments.

In summary, in this work we identified synovial B cell clones diversified within RA synovial ELS which react against RA-FLS-derived autoantigens and characterised RA057/11.89.1 as a novel monoclonal antibody targeting stromal-derived CRT. These results, linked with recent data reporting a high prevalence of anti-CRT antibodies in RA patients suggest that CRT can act as a locally released autoantigen which can be targeted by autoreactive B cells.