Background
A person’s gender plays a major role in the development of rheumatoid arthritis (RA). Nearly 75% of patients with RA are women. The reason for the gender imbalance is unclear, but sex hormones are considered to be of pivotal importance. Particularly, the decrease of estrogen in menopause coincides with an increased risk of developing RA [
1]. Despite this remarkable association, studies addressing the role of estrogen in the development of RA are scarce [
2], and mechanistic studies are virtually absent. Hence, the reason for the preponderance of RA in postmenopausal women remains unclear to date.
RA starts with an autoimmune phase followed by an inflammatory phase [
3‐
5]. Whereas autoimmunity remains clinically silent, inflammation unequivocally leads to symptoms such as pain and swelling. Autoantibodies such as anti-citrullinated peptide antibodies (ACPAs) have a diagnostic, predictive, and prognostic role in RA and can be detected in the preclinical phase several years before the onset of symptoms [
6,
7]. These observations indicate that B-cell-mediated autoimmunity and autoantibody development is crucial for the onset of inflammation in RA. Data derived from mouse arthritis models support this concept by showing that B cells, autoantibodies, and Fcγ receptors (FcγRs), which mediate the effector function of autoantibodies, are necessary for the development of arthritis [
8‐
10].
Besides their role in antigen recognition, antibodies regulate effector cell activation through their constant Fc regions, which bind to FcγRs and activate monocytes/macrophages. Antibodies bear one or several carbohydrate chains, or glycans. Glycan at the Asn297 position at the Fc part of IgG regulates binding capability to FcγRs [
11‐
13]. This glycan is composed of a conserved heptamer that consists of N-acetylglucosamine and mannose residues, which can be extended by fucose, galactose, and finally sialic acids. The composition of the IgG-Fc glycosylation, in particular without terminal sialic acids, determines effector cell activation and hence the inflammatory properties of antibodies [
14]. Low sialylation of Asn297 enhances the proinflammatory activity [
15‐
19], whereas the attachment of terminal sialic acid residues mediates anti-inflammatory effects [
20]. Importantly, it has been shown that the transition from asymptomatic autoimmunity to RA is associated with a change in the sialylation status of antibodies [
20,
21].
Population studies have revealed that IgG-Fc galactosylation and sialylation are higher in premenopausal women than in men but decrease with age [
22,
23]. During pregnancy, when women are protected from RA, IgG-Fc sialylation as well as galactosylation increase [
24,
25], and then reverse within 3 months postdelivery, when RA risk is higher [
25]. Estrogen has been shown to decrease galactosylation of human IgG in healthy individuals [
26], which may explain the increased risk of RA in postmenopausal women. Whether estrogen influences IgG sialylation has not been investigated yet. In the present study, we show that estrogen influences the presence of sialic acid on the Fc glycan of IgG, both in postmenopausal mice challenged by immunization and in postmenopausal women with RA. This effect is based on estrogen-mediated induction of β-galactoside α2,6-sialyltransferases 1 (St6gal1) in plasmablasts, the enzyme that adds sialic acid residues to IgGs. Hence, decrease in estrogen in menopause creates a proinflammatory state characterized by low IgG sialylation and increased risk of developing arthritis.
Methods
Animals and treatments
Female C57BL/6 mice were kept under standard conditions with standard chow and tap water ad libitum. The study was approved by the ethics committees of the Government of Mittelfranken (Germany) and the University of Gothenburg (Sweden). To avoid confounding endogenous sex hormone effects and to mimic a postmenopausal state, mice were ovariectomized (OVX) or sham-operated at 10 weeks of age. Estrogen (17β-estradiol, E2) treatment was carried out with slow-release subcutaneous pellets (Innovative Research of America, Sarasota, FL, USA) containing E2 (0.83 μg/day) or placebo. Treatment of mice with such doses is known to result in serum E2 levels of approximately 60 pg/ml [
27]. In mice, normal serum levels of E2 vary between 25 and 50 pg/ml in diestrus and between 150 and 200 pg/ml in estrus [
28]. Thus, the dose used in this study resulted in physiological serum E2 levels. Treatment efficiency was confirmed by the weighing of the uterus.
Immunization with ovalbumin
Mice were subcutaneously immunized with 100 μg of ovalbumin (OVA) (Sigma-Aldrich, St. Louis, MO, USA) emulsified in complete Freund’s adjuvant (Sigma-Aldrich). Followed with a booster injection the same way including 100 μg of OVA (Sigma-Aldrich) emulsified in incomplete Freund’s adjuvant after 14 days. Serum was taken before OVA immunization, 10 days after initial immunization (day 22), and 10 days after boost (day 38).
Hormone replacement therapy
Postmenopausal women with RA (
N = 49) aged 45–65 years were included in a 2-year, randomized, single-blind, controlled study [
29]. Patients had active disease with at least two of the following criteria: at least six painful joints, at least three swollen joints, erythrocyte sedimentation rate ≥ 20 mm/h, and C-reactive protein ≥10 mg/L. Patients fulfilled the American Rheumatism Association 1987 criteria for RA [
30]. Women in the hormone replacement therapy (HRT) group were given continuous treatment with 2 mg of E2 plus 1 mg of norethisterone acetate daily. All patients provided informed consent, and the ethics committee at the University of Gothenburg (Sweden) approved the study.
Serum measurements
In the sera from the human HRT study, ACPA were evaluated by enzyme-linked immunosorbent assay (ELISA) (Orgentec Diagnostika, Mainz, Germany). In mice, IgGs were measured using commercially available kits (Bethyl Laboratories, Montgomery, TX, USA). OVA-specific IgG were measured with an in-house ELISA, plates were coated with 100 μg/ml OVA (Sigma-Aldrich), incubated with sera (diluted 1:5000), and detected with horseradish peroxidase (HRP)-conjugated polyclonal rabbit antimouse IgG (Dako/Agilent Technologies, Waldbronn, Germany). For evaluation of the affinity properties of antibodies, potassium thiocyanate (Sigma-Aldrich) was added in various doses. For measurement of sialic acid residues on IgG or OVA-specific IgG, biotinylated Sambucus nigra lectin (Vector Laboratories, Burlingame, CA, USA) and streptavidin-HRP (R&D Systems, Minneapolis, MN, USA) were used for detection.
Isolation of OVA-specific antibodies
OVA-specific antibodies were captured from serum of OVA-immunized mice. Protein G-isolated total IgG was dialyzed in sodium phosphate dibasic and enriched over OVA-coupled Sepharose 4B beads (Sigma-Aldrich), washed with NaCl four times, and eluted with lectin buffer. ELISA confirmed enrichment of OVA-specific IgG.
Mass spectrometric analysis for Fc glycans
For the analysis of Fc glycosylation, the IgG eluates were subjected to tryptic digestion by adding 600 ng of tosyl phenylalanyl chloromethyl ketone-treated trypsin (Merck, Kenilworth, NJ, USA) in 40 μl of ammonium bicarbonate buffer followed by overnight incubation at 37 °C. Digested IgG was separated and analyzed using a Dionex UltiMate 3000 UHPLC system (Thermo Fisher Scientific, Waltham, MA, USA) coupled to a maXis Impact HD quadrupole-time-of-flight mass spectrometer (MS) (Bruker Daltonics, Billerica, MA, USA). Details are described in Additional file
1: Supplementary methods. The quality of mass spectra was evaluated on the basis of total intensities per glycopeptide cluster. Analyte curation was performed using the signal-to-noise ratio, isotopic pattern quality, and observed mass-to-charge ratio (
m/z) deviation as obtained after data (pre-)processing with LacyTools [
31]. Following extraction of tryptic glycopeptides by a C18 solid-phase extraction trap column (Dionex Acclaim PepMap 100; Thermo Fisher Scientific), separation was performed with a Supelco Ascentis Express C18 nano-LC column (Sigma-Aldrich) conditioned at 900 nl/min with 0.1% trifluoroacetic acid (mobile phase A), after which the following gradient of mobile phase A and 95% acetonitrile (mobile phase B) was applied: 0 minutes 3% B, 2 minutes 6% B, 4.5 minutes 18% B, 5 minutes 30% B, 7 minutes 30% B, 8 minutes 1% B, and 11 minutes 1% B. The ultraperformance LC was interfaced to the MS with a CaptiveSpray electrospray ionization source and nanoBooster (Bruker Daltonics). Mass spectra were recorded from
m/z 550 to 1800 at a frequency of 1 Hz. Quadrupole ion energy and collision energy of the MS were set at 2 and 5 eV, respectively. The total analysis time per sample was 13 minutes. Detailed calculations are provided in Additional file
1: Supplementary methods.
Cell preparation and flow cytometry
Cell suspensions were obtained from spleen and bone marrow and stained for surface markers after erythrolysis, fixation, and permeabilization (eBioscience, San Diego, CA, USA). Analyses were performed using a Gallios flow cytometer (Beckman Coulter Life Sciences, Indianapolis, IN, USA) and Kaluza Flow Analysis software (Beckman Coulter Life Sciences). The following fluorochrome- or biotin-conjugated antimouse antibodies and reagents were used: allophycocyanin (APC)-conjugated anti-CD267 (transmembrane activator calcium modulator and cyclophilin ligand interactor), fluorescein isothiocyanate (FITC)-conjugated anti-B220, and FITC-conjugated anti-CD11b (all from eBioscience); phycoerythrin (PE)-conjugated anti-CD138 and PE-cyanine 7 (Cy7) conjugated anti-CD3, Pacific Blue F4/80, PE-Cy7 Ly6G, and PE-FcγRIII (CD16) (all from BioLegend, San Diego, CA, USA); APC-conjugated FcγRI (CD64), FcγRIIB, and FcγRIV (self-made); Alexa Fluor 488-conjugated OVA-A488 (Thermo Fisher Scientific); anti-St6gal1) (C) (IBL International/Tecan, Morrisville, NC, USA); and normal rabbit IgG (isotype-matched control antibody) (Thermo Fisher Scientific).
Mouse B-cell proliferation
Splenic B cells were isolated by CD43 depletion using MACS technology (Miltenyi Biotec, Bergisch Gladbach, Germany). The cells were stimulated with lipopolysaccharide (LPS) (Sigma-Aldrich) and cultured for 48 hours to develop them into plasmablasts. The medium was then changed to serum-free medium with no estrogen or 10− 8 M 17β-estradiol (E2) (Sigma-Aldrich) for the last 24 hours. Blocking antibodies toward IL-22 (Poly5164; BioLegend) and tumor necrosis factor (TNF)-α (Ultra-LEAF anti-TNF-α, MP6-XT22; BioLegend) were added. Seventy-two hours after initial seeding and 24 hours after change in medium, supernatants were collected and cells were isolated for RNA analyses.
Plasma cell isolation
CD138+ splenic plasmablasts were isolated using MACS technology (Miltenyi Biotec) from OVX OVA-immunized mice treated with estrogen or placebo. The purified cells were isolated for RNA analyses.
Human B-cell proliferation
Human B cells were purified from peripheral blood mononuclear cells using immunomagnetic beads (Dynal® B Cell Negative Isolation Kit; Thermo Fisher Scientific). The cells were stimulated with TLR-9 agonist cytosine-phosphate-guanine (CpG) oligodeoxynucleotide 2006 (InvivoGen, San Diego, CA, USA), goat anti-human IgA/IgG/IgM F(ab′)2 fragments (Jackson ImmunoResearch Laboratories, West Grove, PA, USA), and human recombinant IL-2 (R&D Systems), then cultured for 5 days to develop them into plasmablasts. After 5 days, medium was changed to no estrogen or 10− 8 M 17β-estradiol (E2) (Sigma-Aldrich). Seven days after initial seeding and 48 hours after medium change, cells were isolated for RNA analyses.
RT-PCR
Total RNA was extracted using an RNeasy kit (Qiagen) and transcribed into complementary DNA using oligo(dT) primers and MuLV reverse transcriptase (Roche Diagnostics, Indianapolis, IN, USA). qRT-PCR was performed with SYBR Green I dTTP (Eurogentec, Liège, Belgium) or the Applied Biosystems StepOnePlus™ Real-Time PCR system (Thermo Fisher Scientific) using an assays-on-demands primer and probe set. The gene expression values were normalized to those of the control gene encoding β-actin and 18S. Primer sequences are described in Additional file
1: Supplementary methods.
Statistical analysis
Statistical analyses were performed using Prism software (GraphPad Software, La Jolla, CA, USA). Two separate groups were compared with unpaired Student’s t tests. One-way analysis of variance followed by Bonferroni multiple-comparisons tests were performed for selected columns. Outliers were eliminated using Grubbs’ test. The correlation was investigated with Spearman’s correlation coefficients. Data are presented as mean ± SEM or as scatterplots, and p < 0.05 was considered significant.
Discussion
In the present study, we show that E2 influences IgG glycosylation, especially the sialylation of IgG, by upregulating the key enzyme St6gal1 in plasmablasts. Deficiency of E2, like in menopause, leads to decreased antibody sialylation and to a proinflammatory IgG pattern, which could influence the onset of RA and may explain the increased risk of RA in postmenopausal women. However, E2 administration increased sialylation of IgG, shifting the antibody effector function to a more regulatory anti-inflammatory pattern, which is supported by a negative correlation between the degree of IgG sialylation and galactosylation with RA disease activity.
In RA, it is well established that autoantibody formation precedes the symptomatic inflammatory phase of the disease. Factors that shift asymptomatic autoimmunity to inflammation are therefore of key interest in understanding the onset of disease [
36]. The sialylation status of autoantibodies seems to play a crucial role in this shift. Low-level IgG sialylation promotes progression to inflammation, whereas high-level sialylation promotes suppression of symptoms [
20]. Our results indicate that estrogen affects the pathogenicity of the antibodies, mainly via regulation of IgG-Fc sialylation. Hence, higher levels of E2 create an anti-inflammatory environment by inducing St6gal1, resulting in a higher degree of antibody sialylation. In accordance, the sharp decrease of estrogens in menopause is supposed to switch this environment to low sialylation and a proinflammatory pattern.
In contrast to the findings regarding sialylation, we did not detect any significant effect of E2 on galactosylation in mice or on the expression of B4galt1, the enzyme mediating galactosylation, in mouse and human antibody-producing cells. Nonetheless, in postmenopausal patients with RA, treatment with E2 increased not only sialylation but also galactosylation of IgG. Similarly, we observed strong correlations of IgG-Fc galactosylation and levels of estrogen in the postmenopausal patients with RA. These findings are in accordance with previous results in healthy individuals showing that E2 regulates galactosylation [
26]. Because galactosylation is a prerequisite for sialylation at the Asn297 site of IgG, an effect of E2 on galactosylation might further strengthen the overall E2-induced glycosylation pattern of human IgG.
E2 effects on B-cell [
37‐
39] and plasma cell differentiation [
40,
41] have been reported previously, but functional consequences on the pattern of plasma cell-mediated antibody production have so far been undetermined. On one hand, regulation of St6gal1 by E2 suggests that the overall effect of E2 on the effector pathways of adaptive immunity is a regulatory one and that loss of E2 induces a proinflammatory environment by altering effector functions of antibodies. On the other hand, E2 did not have any consistent effect on specific antibody levels and affinity, suggesting that the key factor by which E2 regulates inflammatory responses is indeed its influence on IgG glycosylation. Future studies need to be done to test whether estrogen treatment in postmenopausal women stimulates St6gal1 in B cells and plasmablasts.
Conclusions
The data derived from the present study provide a basis for a molecular concept that could explain why susceptibility to RA changes during a woman’s life and specifically increases in menopause. E2 appears to be a protective factor rather than a risk factor in triggering inflammation in arthritis by inhibiting the proinflammatory effector functions of autoantibodies. Higher rates of flares of RA with the decrease of sex hormones after pregnancy [
42], as well as the accumulation of flares in the second low-estrogen phase of the menstrual cycle [
43], also support this concept. Treatment with E2 may therefore have a beneficial effect in some patients with RA, particularly in those with imminent RA displaying autoantibodies and initial symptoms such as pain, with a high risk of progressing to clinical RA.
Acknowledgements
We acknowledge excellent assistance provided by Silke Winkler, Wolfgang Baum, Merja Nurkkula- Karlsson, Katharina Falk, Petra Henning, and Marcus Söderberg. We also thank Holger Bang (Orgentec Diagnostika) for providing ACPA tests used in the project.