Autoimmune rheumatic disease IgG has differential effects upon neutrophil integrin activation that is modulated by the endothelium

The importance of neutrophils in the pathogenesis of autoimmune rheumatic diseases, such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA), is increasingly recognised. Generation of reactive oxygen species (ROS) and release of neutrophil extracellular traps (NETs) by activated neutrophils are both thought to contribute to pathology; although the underlying mechanisms, particularly the effects of IgG autoantibodies upon neutrophil function, are not fully understood. Therefore, we determined whether purified IgG from patients with SLE or RA have differential effects upon neutrophil activation and function. We found that SLE- and RA-IgG both bound human neutrophils but differentially regulated neutrophil function. RA- and SLE-IgG both increased PMA-induced β1 integrin-mediated adhesion to fibronectin, whilst only SLE-IgG enhanced αMβ2 integrin-mediated adhesion to fibrinogen. Interestingly, only SLE-IgG modulated neutrophil adhesion to endothelial cells. Both SLE- and RA-IgG increased ROS generation and DNA externalisation by unstimulated neutrophils. Only SLE-IgG however, drove DNA externalisation following neutrophil activation. Co-culture of neutrophils with resting endothelium prevented IgG-mediated increase of extracellular DNA, but this inhibition was overcome for SLE-IgG when the endothelium was stimulated with TNF-α. This differential pattern of neutrophil activation has implications for understanding SLE and RA pathogenesis and may highlight avenues for development of novel therapeutic strategies.

(ICAM)-1, a ligand that is upregulated on various cell types following inflammation. In contrast, α M β 2 is more promiscuous and recognises a range of ligands including ICAM-1 and fibrinogen, which are upregulated at sites of tissue injury and during active coagulation. Integrin-mediated adhesion, via β 1 and/or β 2 integrins, is vital to neutrophil activation 17 , leading to the production of reactive oxygen species (ROS) and NETosis [18][19][20][21] . Moreover, evidence indicates that integrin inhibition reduces NET release 15,20 . ARDs are generally characterised by immune dysfunction, with many groups exploring the immune differences in patients with SLE and RA. Given the presence of autoantibodies, it is unsurprising to find that defects in B and T cell regulation have been described in both SLE and RA. B cells contribute to pathology not only through antigen presentation, but also by producing autoantibodies 22 . Studies in RA found that autoantibodies stimulate the production of pro-inflammatory cytokines 22 , which promote T cell, B cell and macrophage activation 22,23 . Greater peripheral B cell activation was observed in patients with SLE 24 , with cells being more sensitive to cytokine stimulation 25 , displaying abnormal receptor-mediated signalling 26 and having a greater propensity to undergo epitope spreading 27 .
Th1 cells have conventionally been considered to drive RA pathology, however there is growing interest in other T cell subsets. Th17 cells secrete pro-inflammatory mediators that suppress regulatory T cell (Treg) generation 28,29 . Elevated Th17 and reduced Treg differentiation have been described in RA patients, which promote inflammatory cell phenotypes 30 . In addition, Tregs isolated from RA patients have limited suppressive activity 31,32 , which is attributed to low expression of cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) 32 . This reduced Treg population, with defective suppressive capability, fails to suppress autoreactive T cells 33 . Aberrant T cell activation has also been linked to SLE pathology, with reports documenting abnormal TCR signalling and T cell hyper-responsiveness arising through defects in an array of signalling molecules [34][35][36][37][38][39] .
Macrophages also contribute to RA pathology through pro-inflammatory cytokine production, ROS generation, release of matrix-degrading enzymes, phagocytosis and antigen presentation 40 . Monocytes isolated from SLE patients have elevated expression of activation markers and adhesion molecules [41][42][43][44] . Aberrant cytokine production has also been described in SLE-derived monocytes 25,45,46 , which promote the production of anti-dsDNA autoantibodies by B cells 47,48 .
Studies identified neutrophils in the synovial fluid of RA patients [49][50][51] , with further reports finding increased neutrophil activation [52][53][54] . Neutrophils have also been implicated in RA mouse models [55][56][57] . RA neutrophils are prone to spontaneous NETosis, with elevated NETs being reported in RA serum that correlated with anti-citrullinated protein antibodies (ACPA) titres and markers of inflammation 14 . Adaptive immunity has an established role in SLE pathology, however in more recent years, neutrophils are being increasingly studied. Early studies noted that SLE sera induced neutrophil aggregation and interfered with phagocytosis and degranulation 58 . Uptake of circulating nucleosomes activates neutrophils 59 , which leads to the secretion of antibacterial proteins. Elevated bactericidal proteins have been documented in SLE serum, which correlate with autoantibody titres 60 and disease activity [61][62][63] . NETosis releases dsDNA and inflammatory cytokines, which also correlate with anti-dsDNA antibodies titres 64 .
The presence of circulating autoantibodies in SLE and RA is well-established, and used to aid diagnosis, predict disease progression and monitor response to treatment. Patients with either SLE or RA commonly present with a multitude of autoantibodies, with a large heterogeneity of antigenic targets 65,66 , each believed to confer pathogenic effects. Early reports found that sub-fractions of SLE-IgG were able to activate cultured endothelial cells and promote adhesion of the monocytic U937 cell line 67 . Some groups have specifically examined the effects of IgG upon neutrophil activation, highlighting an increased propensity of patient-derived neutrophils to undergo spontaneous NETosis as well as serum IgG being able to induce NET release 14,[68][69][70] . However, whilst these studies report IgG effects upon an end-stage functional outcome of neutrophil activation (NETosis), there does not appear to be any evidence examining involvement of β 1 and β 2 integrins, which occurs at an early stage of neutrophil activation.
The precise effects of purified ARD-IgG upon neutrophil integrin activation and adhesion are poorly characterised. Given the importance of neutrophil activation to ARD immunopathology, determining the effects of ARD-IgG upon neutrophil function may elucidate the mechanisms through which neutrophils contribute to disease progression. Therefore, we examined the effects of SLE-and RA-IgG on β 1 and β 2 integrin-mediated adhesion and activation.

Results
Patient demographics. Neutrophils were isolated from 12 healthy controls (HCs), 12 SLE patients and 7 RA patients. IgG was purified from a separate cohort of 9 HCs, 5 SLE patients and 9 RA patients. Demographics, clinical history and treatments of all subjects at the time of sample collection are listed in Table 1. All purified IgG samples tested negative for both PR3-and MPO-anti-neutrophil cytoplasmic antibody (ANCA), which excluded any effects being mediated by these traditional anti-neutrophil antibodies. Only RA-IgG samples tested positive (>5 U/ml) for ACPA. Our SLE cohort displayed presented with a variety of manifestations: 59% presented with renal involvement; 29% had arthritis; 12% had pulmonary involvement and 24% presented with photosensitive rashes. All of our RA cohort presented with articular involvement. In addition, 13% were hypertensive, 13% presented with rheumatoid nodules and 25% had pulmonary involvement. All patient fulfilled relevant classification criteria with stable disease. Patients with active disease requiring medication change at the time of sampling were excluded.
Purified SLE-and RA-IgG both enhance hydrogen peroxide production. To  HC-and RA-neutrophils (Fig. 1A). We also found no significant difference between HC-and RA-neutrophils in our assay. To build on these experiments, we subsequently explored the effects of purified IgG upon neutrophils isolated from healthy volunteers, to determine whether the presence of autoantibodies drive the activated neutrophil phenotype reported among patient populations. Here, we found that pre-conditioning neutrophils with SLE-or RA-IgG induced significantly higher rates of H 2 O 2 generation (Fig. 1B), indicating that the presence of autoreactive IgG may prime neutrophils to enhance H 2 O 2 production following stimulation. We therefore found that whilst ex vivo SLE-neutrophils displayed slower rates of H 2 O 2 generation, treatment with ARD-IgG increased H 2 O 2 production rates in HC-neutrophils.

ARD-IgG binds neutrophils.
To determine whether purified IgG bound neutrophils, we employed a flow cytometry approach. HC neutrophils were incubated with pooled IgG (5 different IgG samples per group), chosen at random from each group, and stained for CD15 and human IgG. Neutrophils were then evaluated by flow cytometry, being selected based on their forward and side scatter properties (Fig. 1C). Gated neutrophils were subsequently analysed based on CD15 and the presence of human IgG, both in the absence and presence of FcR blockade (Fig. 1C). Greater numbers of CD15 + IgG + cells were observed with both SLE-and RA-IgG compared to HC-IgG (Fig. 1D). Building on this dataset, we also examined IgG binding in the presence of a FcR blocking reagent (Miltenyi, UK), which only reduced binding of SLE-IgG, indicating a contribution from FcγR-mediated interactions with this IgG source (Fig. 1E). Therefore, all subsequent functional studies were performed with FcγR blockade to specifically examine antigen-specific mediated effects.

ARD-IgG displays differential effects upon neutrophil adhesion and trans-endothelial migration.
To evaluate whether β 1 or β 2 integrin-mediated adhesion were modulated by purified IgG, we evaluated neutrophil adhesion to their respective ligands, fibronectin (via β 1 integrins) and fibrinogen (α M β 2 integrin) in the presence of individual SLE-(n = 5), RA-(n = 7) or HC-IgG (n = 6) samples. In the absence of phorbol 12-myristate 13-acetate (PMA), the effects of SLE-and RA-IgG upon neutrophil adhesion to both fibronectin and fibrinogen were identical to those of HC-IgG ( Fig. 2A,B). In contrast, following PMA stimulation, both SLE-and RA-IgG increased neutrophil adhesion to fibronectin ( Fig. 2A) compared to HC-IgG. Interestingly, only SLE-IgG significantly increased neutrophil adhesion to immobilised fibrinogen (Fig. 2B). The addition of pan-β 1 integrin (P5D2) or α M β 2 -specific (2LPM19c) blocking antibodies reduced PMA-induced binding to values similar to unstimulated cells ( Fig. 2A,B). Thus, PMA-stimulated neutrophil adhesion to fibronectin was enhanced by both SLE and RA-IgG, but to fibrinogen by SLE-IgG only. We then examined the modulatory effects of individual IgG samples upon unstimulated and PMA-induced neutrophil adhesion to human umbilical cord endothelial cells (HUVEC). Interestingly, only SLE-IgG significantly enhanced both unstimulated and PMA-induced neutrophil adhesion to both resting and TNF-α-activated HUVEC compared to control IgG (Fig. 3A,B). In contrast, we found that neither SLE-nor RA-IgG affected basal or chemokine-induced neutrophil trans-endothelial migration compared to HC-IgG (Fig. 3C). Taken together, our data suggest that whilst RA-and SLE-IgG both enhanced PMA stimulated β 1 integrin-mediated adhesion to fibronectin, only SLE-IgG significantly affected α M β 2 -mediated adhesion to fibrinogen as well as cultured HUVEC.

Healthy Control SLE RA
Cohort size (n) 21 17 16 Age (years ± SD) 33 ± 8.1 38 ± 6.8  ARD-IgG increases DNA release by neutrophils. We next examined whether IgG could also modulate the release of extracellular DNA, both in the absence (spontaneous) and presence of PMA (inducible). Using immunofluorescence staining for histone H3, we noted that PMA stimulation induced DNA externalisation, characteristic of NET release (Fig. 4A). Comparative analysis of neutrophil supernatants also showed a significant increase in cell-free DNA in PMA-stimulated cells (Fig. 4B). We therefore used measurements of cell-free DNA, to quantitatively measure DNA released by neutrophils, as an indirect measure of NETosis. We then normalised our results against a no IgG control, thus examining any modulatory effects of IgG upon NETosis. Concurrent with published studies, we also report an increase in spontaneous DNA release by neutrophils in the presence of SLE-or RA-IgG compared to HC-IgG (Fig. 4C). Only SLE-IgG however, significantly elevated PMA-induced DNA externalisation by neutrophils in the absence of endothelial cells (Fig. 4C).
We then studied neutrophil DNA release in co-culture with either steady-state or TNF-α-activated HUVEC to model the cellular interactions that would occur with the vascular endothelium. Interestingly, co-culture of neutrophils with steady-state HUVEC abrogated the increases in both spontaneous and PMA-induced DNA release that we observed with neutrophils stimulated in the absence of endothelial cells (Fig. 4D). However, when HUVEC were activated with TNF-α prior to co-culture, SLE-IgG, but not RA-IgG, significantly increased both spontaneous and PMA-induced DNA externalisation (Fig. 4E).

Discussion
Abnormalities in neutrophil function contribute to the immunopathology of several ARDs, including SLE and RA. Whilst neutrophils isolated from patients with SLE and RA have been shown to exhibit increased propensities to release NETs, it is not known whether pathogenic autoantibodies directly engage and drive neutrophil dysfunction. We report that whole SLE-and RA-IgG fractions have differential binding to neutrophils in vitro, and further demonstrate that ARD-IgG enhance ROS production and DNA release, suggestive of enhanced NETosis. Taken together, these data suggest that ARD-IgG may act as priming agents that enhance neutrophil activation following stimulus.
Many groups have sought to identify the surface molecules implicated in IgG-mediated leukocyte dysfunction. Due to the wealth of potential autoantigens associated with both SLE and RA, identification of a singular antigen target has been met with varying degrees of success and is likely to be completely different in both disease groups. We initially assessed the role of FcγRs, given that immune complexes can interact with neutrophil FcγRs 71 and induce ROS generation 72 . Indeed, SLE-IgG binding was reduced following FcγR blockade, indicating a degree of FcγR-mediated binding. This reduction may indicate a contribution of FcγR-mediated effects to neutrophil activation in patients with SLE. In our further experiments however, we assessed the antigen-specific effects of purified IgG in the presence of FcγR blockade, we found that both SLE-and RA-IgG were able to enhance β 1 integrin activation, measured by adhesion to immobilised fibronectin, but only SLE-IgG significantly enhanced α M β 2 -mediated adhesion to PMA stimulated neutrophils. These differences in integrin activation may highlight a subtle form of differential regulation by different ARD-IgG upon steady-state and PMA stimulated neutrophils. In this model, it is possible that the presence of ARD-IgG is able to enhance integrin affinity and/or avidity further than that seen with PMA alone. This enhancement may allow for different integrin families to either adopt a high affinity conformation or undergo integrin clustering and therefore allow high avidity interactions with integrin ligands. Our experiments utilised immobilised fibronectin and fibrinogen, which are found in areas of tissue injury with local activation of coagulation and in regions of vascular endothelial damage where the sub-endothelial basement membrane has been exposed. Our choice of ligands has the advantage of readily allowing us to distinguish the effects of IgG on β 1 -mediated adhesion to ECM/VCAM-1 from the effects on α M β 2 -mediated adhesion. Given the wealth of literature on ligand binding motifs, it is reasonable to extrapolate our findings to other integrin ligands for α M β 2 and α 4 β 1, such as ICAM-1 and VCAM-1 that play a key role in neutrophil-endothelial interactions. Future work will be aimed at dissecting these ligand preferences further. It is also possible that augmented integrin-mediated binding occurs within the circulation. Fibronectin can be found incorporated into circulating NETs that can promote further neutrophil interactions 73,74 , circulating in plasma, as well as forming a constituent of basement membranes and ECM 75 . In addition, the finding of increased circulating apoptotic endothelial cells in patients with SLE 76 , would provide further endothelial integrin ligands to engage circulating leukocytes within the blood flow.
We found a similar pattern of differential activation with ROS generation, a process requiring α M β 2 77 , that was increased by RA-and SLE-IgG. The requirement of α M β 2 , whose affinity is commonly reduced in SLE 78 , may underlie our observation of reduced ROS in ex vivo SLE-derived neutrophils. Differences in treatment regimens between groups may also account for the reduced rates of H 2 O 2 generation. If we consider both the myelosuppressive and immunosuppressive effects of mycophenolate mofetil and prednisolone, as well as their half-lives (17.6 and 2.6 hours respectively) 79,80 , which all SLE patients were prescribed, it is possible that the activity of freshly isolated neutrophils may still be pharmacologically suppressed. Alternatively, reduced H 2 O 2 production may be a result of metabolic exhaustion arising from systemic activation prior to cell isolation. Integrin activation is central to leukocyte migration, therefore we examined whether ARD-IgG affected neutrophil adhesion to, and migration across, the endothelium. Interestingly, SLE-IgG, but not RA-IgG, significantly augmented neutrophil adhesion to HUVEC, while migration across an IL-8 gradient was unaffected by ARD-IgG. This observation may suggest that the primary effects of ARD-IgG centre on the firm adhesion stage of leukocyte extravasation, as opposed to enhancing both adhesion and trans-endothelial migration. Interestingly, we also noted differences in unstimulated adhesion, with no significant differences observed to immobilised fibronectin or fibrinogen but increased baseline adhesion to HUVEC in the presence of SLE-IgG. This difference may be explained by the presence of numerous additional adhesion molecules on the endothelial monolayer, which are absent in immobilised integrin ligand. For examples, endothelial cells would also express ICAM-1 and ICAM-2, which are ligands for both α L β 2 and α M β 2 on the neutrophil. These differences highlight the complex interactions occurring between integrins and adhesion molecules under physiological conditions. Endothelial co-culture modulated the ARD-IgG effects on NETosis. In particular, the presence of a quiescent endothelium abrogated the effects of SLE-and RA-IgG on DNA externalisation observed in the absence of endothelial cells. Stimulation of the endothelium however, restored elevated SLE-IgG induced NETosis, implicating an endothelial contribution. These results may indicate that the presence of ARD-IgG may enhance neutrophil interactions with their external environment and promote neutrophil activation to regions of endothelial activation and dysfunction. A limitation to these experiments is the use of cell-free DNA as an indirect measurement of NETosis. Whilst our approach allowed us to quantitatively compare DNA externalisation to assess the effects upon NETosis, we lose the ability to specifically assess for markers widely accepted to be associated with NET structure, such as citrullinated histones and MPO. The development of a quantitative NET standard to allow the precise measurements of DNA-protein constructs accepted as NETs would resolve this issue.
We administered FcR blockade in our experiments to focus upon the antigen-specific nature of our observations. To further study this interaction additional experiments could be performed using F(ab′) 2 fragments to ensure cellular effects were mediated solely through antigen-specific binding. Differences in IgG post-translational modifications of the Fc region may also contribute to the physiological effects of IgG. In particular, modifications such as glycosylation and sialylation are believed to mediate cellular effects of IgG 81-85 , and may potentially account for differences between ARD-and HC-IgG. Future experiments however, are required to explore the contribution of any differences in post-translational modifications between IgG groups and were beyond the scope of this study.
These different patterns of IgG-mediated integrin activation and neutrophil function may help elucidate the differential mechanisms driving vascular damage and dysfunction in these different ARDs. In conclusion, we found a differential pattern of autoantibody binding and integrin activation in steady-state and stimulated neutrophils exposed to purified whole IgG from patients with SLE or RA, which was modulated by the activation state of the endothelium (Fig. 5). Further work is now underway to explore the translational implications of our findings to highlight novel therapeutic targets.  IgG purification. Whole IgG was purified by passing patient or HC serum through protein G agarose spin columns (Thermo Scientific, UK) and eluted using 0.1 M glycine (Sigma, UK). Eluted IgG was concentrated, washed and dialysed into endotoxin-free PBS by sequential centrifugations for 20 minutes at 2050 g using centrifugal filter units (Merck Millipore, Ireland). Concentrated IgG was then passed through endotoxin removal columns (Thermo Scientific, UK) and confirmed to be endotoxin-free by EndoLISA (Hyglos, Germany) (<0.06 endotoxin units/ml).

patients.
Autoimmune rheumatic disease-associated IgG binding characterisation. Titres of ACPA in purified IgG samples were quantified using the EDIA anti-CCP-2 kit (Euro Diagnostica, Sweden). Titres of PR3-ANCA and MPO-ANCA in purified IgG were determined using the WIESLAB ANCA screen kit (Euro Diagnostica, Sweden). Quantification of anti-nuclear antibodies (ANA), anti-double stranded DNA antibodies (anti-dsDNA) and rheumatoid factor (RF) were all performed in the University College London Hospital laboratory.
Neutrophil isolation. Citrated  Immunofluorescence staining for neutrophil extracellular traps. Glass coverslips (Fischer, UK) were sterilized with and coated with 200 μg/ml fibrinogen (Sigma, UK) at 4 °C. 5 × 10 5 neutrophils were then added to coverslips and stimulated for 4 hours, after which they were fixed for 15 minutes with 4% PFA at room temperature. Coverslips were blocked in a blocking solution (10% goat serum/1% BSA/2 mM EDTA/HBSS/0.1% Tween-2) overnight at 4 °C, then incubated with 1 µg/ml anti-histone H3 antibody (ab1791, Abcam, UK) and then 2 µg/ml Alexa Fluor 488-conjugated goat anti-rabbit IgG secondary antibody (Life Technologies, UK), which were diluted in blocking solution, for one hour each at room temperature. After incubations, coverslips were washed with HBSS twice, mounted and sealed on microscope slides with ProLong Gold antifade mountant with DAPI (Invitrogen, UK). Slides were subsequently visualized using a Zeiss Axio Imager.A1 inverted fluorescence microscope (Zeiss, Germany) and images analysed using Image J.
Extracellular DNA quantification. HC-neutrophils were pre-treated with IgG for 1 hour, after which NETosis was induced by stimulation with 40 nM PMA for 4 hours. Neutrophil supernatants were obtained by centrifugation at 300 g for 5 minutes and then assessed for extracellular DNA using the Quanti-iT PicoGreen dsDNA kit (Invitrogen, UK). Results were then normalised to a no IgG control, thus examining any modulatory effects of IgG.
Statistical analysis. Data were evaluated using GraphPad Prism. Data were first tested for normality using a Kolmogorov-Smirnov test. In experimental data sets only comparing two groups, a Mann-Whitney test was performed. Where multiple groups were being compared, a Kruskal-Wallis test with a Dunn's multiple comparison test was used. In data sets with two variables, data were assessed by two-way ANOVA with a Dunnet's multiple comparison test. A p value below 0.05 was considered significant.