Sialylated Autoantigen-Reactive IgG Antibodies Attenuate Disease Development in Autoimmune Mouse Models of Lupus Nephritis and Rheumatoid Arthritis

Pro- and anti-inflammatory effector functions of IgG antibodies (Abs) depend on their subclass and Fc glycosylation pattern. Accumulation of non-galactosylated (agalactosylated; G0) IgG Abs in the serum of rheumatoid arthritis and systemic lupus erythematosus (SLE) patients reflects severity of the diseases. In contrast, sialylated IgG Abs are responsible for anti-inflammatory effects of the intravenous immunoglobulin (pooled human serum IgG from healthy donors), administered in high doses (2 g/kg) to treat autoimmune patients. However, whether low amounts of sialylated autoantigen-reactive IgG Abs can also inhibit autoimmune diseases is hardly investigated. Here, we explore whether sialylated autoantigen-reactive IgG Abs can inhibit autoimmune pathology in different mouse models. We found that sialylated IgG auto-Abs fail to induce inflammation and lupus nephritis in a B cell receptor (BCR) transgenic lupus model, but instead are associated with lower frequencies of pathogenic Th1, Th17 and B cell responses. In accordance, the transfer of small amounts of immune complexes containing sialylated IgG Abs was sufficient to attenuate the development of nephritis. We further showed that administration of sialylated collagen type II (Col II)-specific IgG Abs attenuated the disease symptoms in a model of Col II-induced arthritis and reduced pathogenic Th17 cell and autoantigen-specific IgG Ab responses. We conclude that sialylated autoantigen-specific IgG Abs may represent a promising tool for treating pathogenic T and B cell immune responses in autoimmune diseases.

The ability of IgG antibodies (Abs) to modulate immune responses depends on the Ab subclass and the structure of the N-glycan attached to Asn-297 in the Fc region that affect IgG binding to activating and inhibitory Fcγ receptors (FcγRs) on effector cells (1,2). The biantennary core of the Fc glycan consists of four N-acetylglucosamines (GlcNAcs) and three mannoses, which can be further modified with fucose, bisecting GlcNAc, galactose and terminal sialic acid residues ( Figure S1 in Supplementary Material). The abundance of non-galactosylated (agalctosylated; G0) serum IgG Abs that lack galactose and terminal sialic acid residues positively correlates with the disease severity in rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE) , whereas alleviated disease activity in RA patients during pregnancy or after anti-TNF treatment is associated with increased levels of sialylated IgG Ab (6,17,20,(26)(27)(28). Intriguingly, this correlation is especially prominent when only autoreactive IgG Abs are analyzed (22), suggesting that G0 IgG Abs may exacerbate autoimmune inflammation in an antigen-specific manner. Indeed, agalactosylated, but not sialylated, IgG autoantibodies (autoAbs) are able to induce disease symptoms in passive models of arthritis (9,24).
With regard to the development of differently Fc glycosylated IgG Abs, it has been shown that immune responses under inflammatory conditions induce plasma cells (PCs) that generate G0 IgG, whereas immune responses under tolerogenic conditions induce more galactosylated and sialylated IgG Abs (29)(30)(31)(32).
The anti-inflammatory effects of sialylated IgG Abs have first been reported for the intravenous immunoglobulin (IVIG)pooled human serum IgG from healthy donors (33)(34)(35). The sialylated IVIG fraction attenuates arthritis in mice via its binding to the C-type lectin receptor SIGN-R1 (specific ICAM-3 grabbing non-integrin-related 1) on regulatory marginal-zone macrophages (36), and thereby induces an anti-inflammatory environment and upregulates the inhibitory Fcγ receptor FcγRIIB on effector macrophages (37). Moreover, sialylated IVIG is able to inhibit dendritic cell (DC) maturation through an FcγRIIBindependent mechanism (29,(38)(39)(40). Together, these data suggest that the sialylated IVIG fraction exerts anti-inflammatory effects on both innate and adaptive immune cells.
Under physiological conditions, IgG Abs mediate their effector functions through the formation of immune complexes (ICs) with an antigen (29)(30)(31)(32)41). Recent reports suggest that sialylation of antigen-specific IgG Abs affects their effector functions and the course of an immune response (29,30). In the context of autoimmunity, application of small amounts of sialylated IgG autoAbs has reduced joint swelling in the collagen-induced arthritis (CIA) model (24). Furthermore, endogenous sialylation of IgG Abs have attenuated disease development in mouse models of nephritis and arthritis through a pathway similar to IVIG (42). Finally, ICs containing sialylated antigen-specific IgG Abs have inhibited LPS-induced IL-6 production by DCs in vitro (29).
To further investigate the protective effect of sialylated autoantigen-specific IgG Abs on the development of autoimmune pathology, here we studied the disease course in lupus nephritis-prone FcγRIIB-deficient (Fcgr2b −/− ) mice and in 56R +/− Fcgr2b −/− mice that express a transgenic self-and polyreactive B cell receptor (BCR) (43)(44)(45)(46) and produce T cellindependent sialylated IgG2a and IgG2b autoAbs (47). We further tested how sialylated collagen type II (Col II)-reactive monoclonal murine IgG Abs influence the development of Col II-induced arthritis (CIA), accumulation of Th1 and Th17 cells, and autoAb production. Our results suppose that sialylated IgG autoAbs attenuate the development of pathogenic autoimmune conditions and might affect inflammatory T and B cell responses.

Depletion of cD4 + T cells
For depletion of CD4 + T cells, mice were injected intraperitoneally (i.p.) with 250 µg of anti-mouse CD4 (GK1.5) every 4 days for the indicated period of time. GK1.5 hybridoma Abs were purified from hybridoma cultures using protein G Sepharose. The depletion of CD4 + T cells (blood samples) was verified via flow cytometry ( Figure S2 in Supplementary Material). sialylation analysis of serum igg abs From wt, Fcgr2b −/− and 56r +/− Fcgr2b −/− Mice Serum IgG Abs from the indicated wt, Fcgr2b −/− and 56R +/− Fcgr2b −/− mice were purified using protein G Sepharose. To characterize the sialylation of purified IgG Abs, the GlykoScreen™ Sialic Acid Quantification Kit (Prozyme, Hayward, CA, USA) was utilized according to the manufacturer's instructions. In brief, sialic acid molecules were enzymatically released from the purified IgG Abs by incubation with sialidase A for 2 h at 37°C. Released sialic acid molecules were enzymatically converted in a two-step process to acetylphosphate and hydrogen peroxide. Addition of HRP catalyzed a reaction of hydrogen peroxide with another added substrate into a fluorescent dye, which was quantified at 590 nm.

In Vitro De-sialylation of igg abs
De-sialylation of purified polyreactive serum IgG Abs was performed with the Prozyme Sialidase kit (Prozyme).
glycan analysis of Polyreactive serum igg abs From 56r +/− Fcgr2b −/− Mice and Monoclonal igg abs via MalDi-TOF Ms N-glycans were isolated from purified IgG samples via hydrolysis with recombinantly expressed endoglycosidase S (EndoS) from Streptococcus pyogenes (54). EndoS cleaves the Fc N-glycans of IgG Abs between the first and second GlcNAc ( Figure S1 in Supplementary Material). The resulting N-glycans were purified through solid phase extraction using reversed-phase C18 and graphitized carbon columns (Alltech, Deerfield, IL, USA). The samples were then permethylated according to standard protocols (30,55) and further investigated via MALDI-TOF MS in duplicate. The spectra were recorded on an Ultraflex III mass spectrometer (Bruker Daltonics Corporation, Billerica, MA, USA) equipped with a Smartbeam laser. Calibration was performed on a glucose ladder, and 2,5-dihydroxybenzoic acid was used as the matrix. Spectra were recorded in reflector positive ionization mode, and mass spectra from 3,000 laser shots were accumulated. Based on the terminal sugar moiety, the EndoS resulting peaks were assigned to one of the following nine groups: G0+ bisecting GlcNAc, G0 w/o bisecting GlcNAc, G1+ bisecting GlcNAc, G1 w/o bisecting GlcNAc, G2+ bisecting GlcNAc, G2 w/o bisecting GlcNAc, G1S1, G2S1 and G2S2 ( Figure S1 in Supplementary Material). Peaks containing both sialic acid and bisecting GlcNAc were not detected. In general, murine IgG Abs hardly showed bisecting GlcNAc structures. However, the calculated proportions of the bisecting GlcNAc versions of G0, G1 and G2 were added to the percentages of the G0, G1 and G2 versions without bisecting GlcNAc, respectively, to get six groups totaling 100%: G0, G1, G2, G1S1, G2S1 and G2S2. In some figures the percentages of S1 (G1S1 + G2S1) and S2 (G2S2) glycans are presented.

nephrotoxic nephritis-induced Mouse Model
Nephritis was induced by injection of 100 µg of sheep IgG Abs in CFA on day 0, followed by intravenous (i.v.) injection of 80 µl of sheep anti-glomerular basement membrane (anti-GBM) nephrotoxic serum (NTS) 4 days later (56). Development of nephritis was verified by the detection of proteinuria as described above.
Col II-reactive monoclonal IgG1 Abs were produced via polyethylenimine (PEI; Sigma-Aldrich)-mediated cotransfection of human embryonic kidney 293 cells with plasmid DNA encoding the IgH and IgL chains ( Figure S4 in Supplementary Material). IgG Ab integrity was analyzed through SDS gel electrophoresis, while Ab reactivity was controlled via ELISA, and IgG Fc N-glycosylation was analyzed using MALDI-TOF MS.

chicken cOl2-induced arthritis (cia) Mouse Model
Chicken type II collagen (Col II; Sigma-Aldrich) was dissolved at 2 mg/ml in 0.05 M acetic acid and emulsified in an equal volume of enriched CFA (see reagents). Then, 8-10-week-old Fcgr2b −/− mice were immunized subcutaneously (s.c.) with 100 µl of the emulsion (equivalent to 100 µg of Col II). On day 21, a booster s.c. injection of 100 µg of chicken Col II in IFA was administered. Mice were monitored for swelling encompassing the paw and ankle or ankylosis of the limb to determine the onset and severity of the disease in a blinded manner. The swelling of each foot was scored as follows: healthy paws and ankles (score 0) showed no abnormal swelling, redness, contact sensitivity or motor activity alterations. Low swelling of paws and/or ankles was scored with 1, prono unced swelling with 2 and severe balloon-like whole swelling (ankylosis) with 3; thus, each mouse could achieve a maximum score of 12. The mean clinical score was calculated by totaling the scores of all mice in a group and dividing by the number of mice in that group. Prevalence indicates the percentage of animals in an individual group with a score >0 on the indicated time point. Onset of disease was specified at the indicated day, at which an animal reached score >0 for the first time.

ankle histology
Ankle samples were embedded in paraffin and sections were stained with hematoxylin and eosin (H&E) or anti-CD3. Immunofluorescence was detected using a Leica DM IRE confocal laser scanning microscope.

statistical analysis
Statistical analyses unless otherwise stated, were performed using Student's t-test comparing two groups or One-way ANOVA for more groups, respectively, or the logrank test for survival curves: *P < 0.05, **P < 0.01, and ***P < 0.001. If not stated otherwise, murine data were taken from one representative out of 2-5 individual experiments or combined from multiple experiments and are presented as the mean or median (median fluorescence intensity) values as indicated ±SEM; each data point represents an individual animal.
Fcgr2b −/− mice positive for IgG autoAbs started to develop proteinuria by the age of 6 months, and about a half of the Fcgr2b −/− mice died due to severe nephritis by the age of 9 months (Figures 1A,B and data not shown) (46,49,50). Mice developing nephritis showed, in addition to IgG Ab depositions, also macrophage infiltrations in the kidney ( Figure 1A). Inversely, mice with IgG Ab depositions in the kidney, but without macrophage infiltrations, did not show proteinuria and nephritis ( Figure 1A).
Earlier studies have shown that 56R +/− mice on the C57BL/6 wt background produce autoreactive IgM and some IgG Abs (47), and that the introduction of the 56R allele into lupus-prone Fcgr2b −/− mice lead to increased IgG class switched autoAbs, particular of the 56R allele (45,46). Whereas the generation of IgG2a and IgG2b autoAbs in Fcgr2b −/− mice requires T cell help (50,60), autoAbs in 56R +/− wt mice largely develop in a T-cellindependent manner (47). Because we recently demonstrated that T cell-independent B cell activation induces immunosuppressive sialylated IgG Abs in vivo (30), we wondered whether the introduction of the 56R allele into lupus-prone Fcgr2b −/− mice may lead to T cell-independent IgG autoAbs and provide a disease-protective effect.
In line with previous reports, 56R +/− Fcgr2b −/− mice generated high serum titers of class-switch DNA-, nucleosome-and polyreactive IgG2aa and IgG2b Abs, which formed depositions in the kidney (Figure 1; Figure S2 in Supplementary Material) (45,46). In contrast to Fcgr2b −/− mice, all 56R +/− Fcgr2b −/− mice developed IgG2aa and IgG2b autoAbs already by the age of 2 months (45,46), which (IgG2aa, but not IgG2b) only slightly further increased until the age of 6 months ( Figure 1E). We also found comparable anti-nuclear reactivity of IgG2a Abs in the sera of 5-7 months old Fcgr2b −/− and 56R +/− Fcgr2b −/− mice ( Figure 1F). However, despite the early presence of IgG Abs of similar antigen specificity and subclass, and IgG Ab deposition in the kidney, none of the 56R +/− Fcgr2b −/− mice showed macrophage infiltration into the kidney and proteinuria by the age of 9 months (Figures 1A,B).  (Figures 1A and  2A,B). The subsequent development of lupus nephritis (manifested by proteinuria) was additionally associated with further enhanced splenomegaly, increased frequencies of Th17 cells and infiltration of macrophages into the kidneys (Figures 1A,B and  2A,B). These findings confirm recent studies showing that the IL-17 signaling pathway is important for the development of disease in lupus-prone mice (50).
In contrast, 56R +/− Fcgr2b −/− mice showed no signs of autoimmune inflammation (Figures 2A,B). Together these findings suggest that 56R-derived IgG autoAbs may be able to actively protect lupus-prone FcγRIIB-deficient mice from developing autoimmune inflammation.
Autoimmune-prone MRL-Fas(lpr) mice and SLE patients show increased levels of pro-inflammatory agalactosylated (G0) IgGs in serum, compared to healthy controls (8,12,15). In line with that, serum IgG Abs from lupus-prone Fcgr2b −/− mice were less sialylated, compared to wild-type controls (Figures 3C,D; Figure  S1 in Supplementary Material). Reduced IgG sialylation was especially evident in Fcgr2b −/− mice that developed signs of lupus nephritis (proteinuria and kidney inflammation) (Figure 3D).  In contrast, serum IgG sialylation in 56R +/− Fcgr2b −/− mice was comparable to that of healthy wild-type animals ( Figure 3D). Notably, the protein expression level of the alpha2,6sialyltransferase 1 (St6gal1), which is responsible for terminal sialylation of IgG Fc glycan (24,(29)(30)(31), was reduced in total and IgG-switched PCs (IgG + PC) of Fcgr2b −/− mice with nephritis, but not in PCs of 56R +/− Fcgr2b −/− mice (Figures 3E,F). These data are consistent with our recent findings that T cell independent B cell activation leads to the development of PCs expressing high levels of St6gal1 and producing sialylated IgG Abs (30).

Transfer of ics containing sialylated igg abs inhibit nephritis in Fcgr2b −/− Mice
To address such a possible ameliorating effect, we next tested whether ICs containing sialylated IgG autoAbs from 56R +/− Fcgr2b −/− mice can directly attenuate the onset of nephritis in an induced nephritis model (53). We purified polyreactive TNP-binding ( Figure S2 in Supplementary Material) IgG Abs from sera of 56R +/− Fcgr2b −/− mice and generated ICs with TNPcoupled sheep (TNP-sheep) IgG Abs using either the purified native (sialylated) or in vitro sialidase-treated (de-sialylated) polyreactive IgG Abs (Figures 4A,B). The ICs were then transferred to Fcgr2b −/− mice and, 2 weeks later, the nephritis was induced by injecting TNP-sheep IgG Abs in CFA followed by i.v. injection of sheep anti-GBM NTS 4 days later (Figure 4) (56).
The ICs containing de-sialylated polyreactive IgG Abs increased nephritis-induced mortality when compared to a positive control ( Figure 4C). In contrast, the ICs containing native sialylated polyreactive IgG Abs attenuated nephritis-induced mortality in Fcgr2b-/-mice ( Figure 4C). Similarly, ICs containing in vitro sialylated anti-TNP monoclonal murine IgG1 (clone H5) Abs, but neither native non-sialylated anti-TNP monoclonal nor sialylated antigen-unspecific monoclonal murine IgG1 Abs, reduced mortality in this nephritis model (Figures 4D-F; Figure S3 in Supplementary Material). In summary, these results showed that only antigen-specific sialylated IgG Abs were able to attenuate disease development.

sialylated collagen-specific igg autoabs attenuate autoimmune inflammation in the cia Model independent of FcγriiB
In order to see whether these sialylation dependent attenuating effects are detectable in a broader spectrum of autoimmune disease models, we further analyzed whether sialylated IgG autoAbs are able to attenuate autoimmune pathology and inflammation also in the collagen type II-induced arthritis (CIA) model. In contrast to earlier studies (24) we chose FcγRIIB-deficient mice for our experiments (66), because the observed inhibitory effect of sialylated IgG autoAbs in the former experiments was FcγRIIBindependent.  We produced two monoclonal murine Col II-reactive murine IgG1 Abs (clones M2139 and CII 1-5; 57, 58) in a native, very low-sialylated form and then generated sialylated forms of these Abs by in vitro galactosylation and sialylation (Figures 5A,B; Figures S1 and S5 in Supplementary Material), which do not affect antigen reactivity ( Figure S5 in Supplementary Material). (e) Fc sialylation of the monoclonal anti-TNP murine IgG 1 Abs (clone H5) before and after in vitro sialylation was determined through EndoS-treatment and MALDI-TOF MS (the percentage of one or two sialic acid residues coupled to the glycan: S1, S2; Figure S1 in Supplementary Material). (F) Kaplan-Meier survival analysis for the indicated groups. One representative experiment out of two independent experiments is shown.  (Figure 5A). 100 µg of the Col II-reactive IgG1 Abs (either the low-sialylated or the sialylated form) were administered twice -1 day before and 9 days after the first immunization ( Figure 5A). Foot swelling (clinical scores of 0-3 per foot with a maximum clinical score of 12 per mouse) was used as the marker to assess the CIA reaction.
The sialylated, but not the low-sialylated, Col II-reactive IgG Abs significantly reduced the mean clinical score of foot swelling, as compared to a PBS-treated control group (Figures 5C,D).
In detail, only about 50% of the mice treated with the sialylated Col II-specific IgG1 Abs started to develop foot swelling (clinical score > 0 per mouse), whereas more than 80% of the mice treated with low-sialylated Col II-specific IgG Abs or with PBS developed foot swelling (Figures 5D,E). No significant differences in the timing of the disease onset were observed between the groups (Figure 5F). Together, these data are consistent with recent findings that sialylated Col II-specific IgG1 Abs can attenuate CIA in DBA/1 mice (24). Their studies further showed that the suppressive effect of sialylated IgG autoAbs was autoantigen-specific; an antigen-unspecific sialylated IgG1 Ab failed to attenuate CIA in their model (24). Here, we further show, that the attenuation of CIA with sialylated IgG autoAbs is independent of FcγRIIB.
Furthermore, the effect of 100 µg of the different sialylated Col II-specific IgG1 Abs was compared to the effect of high (50 mg; approximately 2 g/kg) and low (100 µg; approximately 4 mg/kg) doses of IVIG on the induction of CIA and particular inflammatory T and B cell responses in the CIA model ( Figure S5 in Supplementary Material). We found that high doses of IVIG attenuated the mean clinical score and the prevalence of CIA such as low amounts (1/500 compared to high dose IVIG) of sialylated IgG autoAbs ( Figure S5 in Supplementary Material), whereas administration of equal amounts of non-specific (sialylated) IgG  (low doses of IVIG) was insufficient to alleviate autoimmune inflammation. Random analysis of ankle sections by histology H&E and anti-CD3 staining showed no differences between mice from different groups with identical foot scores ( Figure S5 in Supplementary Material and data not shown). Also, high, but not low, doses of IVIG and sialylated, but not low-sialylated, Col II-reactive IgG autoAbs reduced by trend the accumulation of Th1 cells and significantly the accumulation of inflammatory Th17 cells, which are known for their important role also in the pathogenesis of CIA (Figure 6A) (67)(68)(69)(70)(71). Interestingly, unlike sialylated Col II-specific IgG1 Abs, the high doses of IVIG failed to inhibit the generation of Col II-specific IgG2 autoAbs ( Figure 6B) suggesting a different or additional mechanism of low doses of sialylated Col II-specific IgG Abs as compared to high doses of IVIG.
We could not detect a significant influence of sialylated or lowsialylated anti-Col II IgG Abs on the frequency of total Foxp3 + regulatory CD4 T (Treg) cells in the CIA model (data not shown). However, we observed an increase in antigen-specific Foxp3 + Treg frequencies and a tendency toward a reduction of antigen-specific CD4 T cell proliferation with sialylated IgG Abs as compared to low-sialylated IgG Abs in a transfer model with OVA-specific (OT-II) CD4 T cells in C57BL/6 wt and Fcgr2b −/− mice ( Figure  S6 in Supplementary Material).
Also, the induction of OVA-specific Foxp3 + Tregs by sialylated monoclonal TNP-specific IgG Abs seemed to be antigen-specific, as mice immunized with OVA (in contrast to TNP-OVA) failed to elicit a comparable increase in Foxp3 + Treg frequencies ( Figures  S6D,E

in Supplementary Material).
These data further suggested an effect of sialylated antigenspecific IgG Abs on the adaptive immune response through an FcγRIIB-independent mechanism.
Matured dendritic cells (DCs) that produce inflammatory cytokines are a prerequisite for induction of inflammatory T and B cell responses and there is evidence that ICs containing sialylated IgG Abs can inhibit DC activation (29). We assessed how IgG sialylation modulates the capacity of ICs to suppress IL-6 production, a cytokine critical for Th17 generation (72,73) and CIA development (70,71,74,75), by BM-derived FcγRIIB-deficient DCs in vitro (Figure 6C). IL-6 secretion induced by treating DCs with ICs containing asialylated Col II-reactive IgG1 autoAbs was dramatically reduced by adding sialylated IgG Abs (Figure 6C).
In summary, the data suggest that ICs containing antigen and sialylated antigen-specific IgG Abs can influence DC Collagen-induced arthritis (CIA) was induced in Fcgr2b −/− mice as described in Figure 5 and Figure S5C in Supplementary Material. One day before and 9 days after the first immunization, the mice received 100 µg of either low-sialylated (low-sial; n = 10) or in vitro galactosylated plus sialylated (+sial; n = 9) anti-Col II murine IgG1 Abs (clones M2139 and CII 1-5; 50 µg each) or high dose (50 mg; n = 10) or low dose (100 µg; n = 10) of intravenous immunoglobulin (IVIG) ( Figure S5B

DiscUssiOn
Fc glycosylation of IgG molecules regulates their effector functions and thereby may critically contribute to development of autoimmune pathology. On the one hand, agalactosylated IgG autoAbs are associated with severity of autoimmune disorders, such as RA and SLE , and able to induce disease symptoms in mouse models of RA (9,24). On the other hand, the presence of autoAbs in sera of many healthy humans (76)(77)(78)(79) implies that additional factors, including Fc glycosylation, seem crucial for rendering autoAbs pathogenic. For instance, RA patients develop IgG autoAbs long before clinical symptoms (21,80), and the glycosylation patterns of these early autoAbs are less inflammatory compared to IgG Abs detected in patients with the manifested disease (21,25). Another unexplored possibility is that anti-inflammatory sialylated autoAbs not only lack pathogenic activity, but may be responsible for inducing immune tolerance to autoantigens in healthy animals and humans. Indeed, we have recently demonstrated that T-cell-independent antigens promote generation of immunosuppressive sialylated IgG molecules (30), whereas T-cell-dependent immunizations can induce antigen-specific IgG Abs with pro-or anti-inflammatory glycosylation patterns depending on the co-stimuli (29,30). Our present study further suggests that sialylated IgG2a and IgG2b autoAbs produced by 56R BCR-expressing B cells in a T-cell-independent manner are able to attenuate development of nephritis in lupus-prone FcγRIIB-deficient mice. These findings therefore would extend the existing knowledge that B cells regulate immune responses and inhibit autoimmune pathology via secretion of immunosuppressive cytokines, such as IL-10 and IL-35 (81).
Antigen specificity of sialylated IgG Abs may play a critical role in exerting their anti-inflammatory effect. Indeed, the transfer of ICs containing sialylated polyreactive IgG autoAbs from 56R +/− Fcgr2b −/− mice or sialylated antigen-specific monoclonal IgG Abs attenuated the development of the induced nephritis (Figure 4), whereas antigen-unspecific sialylated monoclonal IgG Abs failed to reach the inhibitory potential of antigen-specific sialylated IgG Abs.
In the arthritis model, sialylated Col II-specific IgG Abs reduced arthritis symptoms (Figure 5), whereas administration of equal amounts of non-specific (sialylated) IgG (low dose of IVIG) was insufficient to alleviate autoimmune inflammation. These data are well consistent with recent reports showing that sialylated Col II-specific IgG1 Abs can inhibit CIA in DBA/1 mice (24). Importantly, Ohmi et al. demonstrated that the inhibitory effect of sialylated IgG autoAbs in the CIA model is autoantigenspecific, since non-specific sialylated IgG1 failed to suppress CIA (24). We cannot exclude, however, that the suppressive effects of sialylated IgG autoAbs observed here are only partially mediated in an antigen specific manner.
Mouse studies that used IVIG and sialylated Fc fragments suggest that antigen specificity is not essential for the antiinflammatory action of the sialylated subfraction of IVIG (35). By comparing the effects of high doses of (sialylated) IVIG and sialylated collagen-specific IgG Abs, our data suggest that antigen specificity might significantly enhance the capacity of sialylated Abs to inhibit immune reactions. In accordance, we found that small amounts of collagen-specific sialylated IgG1 Abs, but not high doses of IVIG, were able to inhibit the development of IgG2 autoAbs in the CIA model. Since antigen specificity is necessary for IC formation, antigen-specific sialylated IgG Abs might inhibit IgG autoAb production via an alternative pathway that potentially requires IC formation.
Extensive evidence suggest that generation of Th17 cells plays a crucial role in pathogenesis of many autoimmune disorders and mouse models of SLE and CIA are dependent on IL-17 (50,67,70,71). Moreover, IL-17 is necessary for development of pathogenic G0 IgG Abs (30). IL-6 skews T cell differentiation toward IL-17Aproducing Th17 cells, suppressing the generation of Foxp3 + Treg cells (72,73). In line, we showed that sialylation of IgG autoAbs reduces IL-6 production by DCs in vitro and Th17 cell accumulation in autoimmune models. Moreover, we observed that only the formation of sialylated IgG ICs increases the frequencies of antigen-specific Foxp3 + Treg cells in the OT-II + T cell transfer model.
In summary, we suppose that ICs containing autoantigenspecific sialylated IgG Abs influence inflammatory DC activation and IL-6 production in a FcγRIIB-independent manner and thereby downregulate Th17 generation, formation of pathogenic G0 autoAbs and, hence, alleviate clinical signs of autoimmune pathology.       Figures 5 and 6 and Figure S5 in Supplementary Material. The start and stop codons are colored in red. The restriction sites used for cloning are highlighted in yellow. The first box (blue) represents the leader sequence. The second box (yellow) shows the variable VDJ or VJ sequence whereas the third box (gray) is the constant IgG1-Fc or kappa chain sequence, respectively. (a) col ii-reactive (clone M2139) murine igg1 heavy chain (accession number: MH208236). The 43 bold and underlined bases behind the leader sequence were missing in the original sequence description (NCBI Z72462) (57) and completed here by the J558.2.88 (NCBI BN000872) sequence because of its highest homology to the original sequence observed after NCBI, IgBlast alignment. The C57BL/6 IgG1 heavy chain constant region starts two bases in front of the SalI restriction site. Because of the introduction of the SalI restriction site, the constant IgG1 heavy chain sequence starts with the amino acids (A)STT… instead of (A)KTT…, which had no functional influence (60). (B) col ii-reactive (clone M2139) murine kappa chain (accession number: MH208237). The first 27 bold and underlined bases behind the leader sequence were missing in the original sequence description (NCBI Z72463) (57) and were completed here by the 21-1 (NCBI X16955) sequence because of its highest homology to the original sequence observed after NCBI, IgBlast alignment. The IGKJ2 sequence in the original sequence description was incomplete and completed here with bold and underlined letters representing the IGKJ2 sequence (accession number V00777) identified on the ImMunoGeneTics (IMGT) Marie-Paule homepage (http://www.imgt.org). The C57BL/6 constant light chain region starts with the BsiWI restriction site. Because of the introduction of the BsiWI restriction site, the constant kappa light chain sequence starts with the amino acids (R)TDA… instead of (R)ADA…, which had no functional influence (60). (c) col ii-reactive (clone cii 1-5) murine igg1 heavy chain (accession number: MH208238). The IGHJ2 sequence in the original sequence description (NCBI MMU69538) (58) was incomplete and completed here with bold and underlined letters representing the IGHJ2 sequence (accession number V00770) identified on the ImMunoGeneTics (IMGT) Marie-Paule homepage (http://www.imgt.org). The C57BL/6 IgG1 heavy chain constant region starts two bases in front of the SalI restriction site. Because of the introduction of the SalI restriction site, the constant IgG heavy chain sequence starts with the amino acids (A)STT… instead of (A)KTT…, which had no functional influence (60). (D) col ii-reactive (clone cii 1-5) murine kappa chain (accession number: MH208239). The four bold and underlined bases behind the leader sequence were corrected from the original sequence description (NCBI MMU69539) (58) because of its obvious mismatching from the highly homologous 12-44 (NCBI AJ235955) sequence observed after NCBI, IgBlast alignment probably resulting by sequencing/analyzing mistakes. The exchanges were from g to c, c to a, c to g and a to g. The IGKJ1 sequence in the original sequence description was incomplete and completed here with bold and underlined letters representing the IGKJ1 sequence (accession number V00777) identified on the ImMunoGeneTics (IMGT) Marie-Paule homepage (http://www.imgt.org). The C57BL/6 constant light chain region starts with the BsiWI restriction site. Because of the introduction of the BsiWI restriction site the constant kappa light chain sequence starts with the amino acids (R)TDA… instead of (R)ADA…, which had no functional influence (60).
FigUre s5 | Low doses of sialylated Col II-reactive monoclonal IgG antibodies (Abs) and high doses of intravenous immunoglobulin (IVIG) attenuate collageninduced arthritis (CIA). These data are parts of the experiments described in Figures 5 and 6. (a) Col II reactivities of native low-sialylated (low-sial) and in vitro galactosylated plus sialylated (+sial) M2139 and CII 1-5 IgG1 Abs and of an antigen-unspecific, TNP-specific murine IgG1 hybridoma Ab (clone H5; negative control) (52) were determined through ELISA. (B) Graphical representation of the experimental strategy used in Figure 6. CIA was induced in Fcgr2b −/− mice as described in Figure 5a. One day before and 9 days after the first immunization the mice received 100 µg of either non-sialylated (low-sial; n = 10) or in vitro galactosylated plus sialylated (+sial; n = 9) anti-Col II murine IgG1 Abs (clones M2139 and CII 1-5; 50 µg each) or high dose (50 mg; n = 10) or low dose (100 µg; n = 10) IVIG. The positive control group received PBS instead of Abs (n = 10). (c) Foot swelling was scored on the indicated days from 0 to 3 per foot resulting in a maximal clinical score of 12 per mouse. The mean clinical score of all mice and the prevalence (percentage of affected animals with a score > 0) are shown for all groups on the indicated days. (D) Representative foot ankle sections of representative mice with CIA score 0 (left; healthy mouse) or score 2 (right; mouse with a swollen ankle) were stained with hematoxylin and eosin (H&E) or anti-CD3. One representative experiment out of two independent experiments is shown. Differences in disease evolution were calculated using two-way ANOVA.