Immunoglobulin E plays an immunoregulatory role in lupus

The sera from 12–16-wk-old FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice was assessed for the presence of increased levels of autoreactive IgE. At this age, FcγRIIB^−/− mice showed no obvious signs of a lupuslike phenotype although low levels of autoantibodies were present in some mice. In contrast, most of the FcγRIIB^−/−.Yaa mice had elevated levels of autoantibodies and organ pathology, including nephritis. We first explored whether these mice made autoreactive IgE by measuring the levels of dsDNA-specific IgE, the most abundant autoreactive IgE found in humans with SLE (Dema et al., 2014). This was compared with the presence or absence of antinuclear antibodies (ANA) as a determinant of disease activity. As seen in lyn^−/− mice (Charles et al., 2010), all tested sera showed increased amounts of dsDNA-IgE relative to WT mice. However, as shown in Fig. 1 a, FcγRIIB^−/−.Yaa mice that trended toward higher levels of ANA-specific Ig’s also showed the highest level of dsDNA-specific IgE relative to those mice that had lower levels of ANA-specific Ig’s, albeit significance was not achieved. Nonetheless, this suggested an association of dsDNA-specific IgE with the high levels of these autoantibodies in mice that developed lupuslike disease. This led us to explore the effect of IgE on disease development. An almost complete penetrance of lupuslike disease with organ pathology occurs at 16 wk for FcγRIIB^−/−.Yaa and at ∼28 wk for FcγRIIB^−/− mice, which results in significant mortality due to renal failure for both strains (Fig. 1 b). At this age, almost all mice were ANA⁺ for both strains. Thus, we chose to work with the respective strains at the indicated ages to assess the role of IgE in the disease. As shown in Fig. 1 b, the median survival of FcγRIIB^−/−.Yaa and FcγRIIB^−/− mice was 4.8 and 7.3 mo, respectively. Crossing of FcγRIIB^−/−.Yaa and FcγRIIB^−/− mice with Igh7^−/− mice increased the median survival to 11 mo for FcγRIIB^−/−.Yaa Igh7^−/− and was not defined for FcγRIIB^−/− Igh7^−/− mice as all the latter mice survived well beyond a 12-mo period. This was accompanied by decreased C3 complement (Fig. 1 c) and IgG immune complex (Fig. 1 d) deposition in the kidney of these mice as well as decreased glomerulonephritis (Fig. 1 e). Kidney function as determined by the ratio of albumin to creatinine in the urine returned to almost normal levels (Fig. 1 f). These striking findings showed that IgE contributes to the development of disease and organ pathologies.

The marked absence of immune complex deposition in the kidney of FcγRIIB^−/−.Yaa Igh7^−/− and FcγRIIB^−/− Igh7^−/− mice led us to explore whether IgE deficiency was generally affecting the production of pathogenic autoantibodies in these mice. As shown in Fig. 2 a, production of anti-dsDNA, anti-ribonucleoprotein (RNP), or anti-histone IgG was significantly decreased in FcγRIIB^−/−.Yaa Igh7^−/− mice and in FcγRIIB^−/− Igh7^−/− mice, and there was a strong trend toward decreased responses as well. Total IgG2b and IgG3 were also decreased whereas IgG1 appeared to be elevated in FcγRIIB^−/− mice regardless of whether they were IgE sufficient or deficient (Fig. 2 b). These changes suggested that the loss of IgE was broadly affecting B cell Ig production rather than narrowly or selectively impinging on autoreactive B and plasma cells. A hint that this might be the case was provided by the marked reversal of the splenomegaly and cellularity seen in the FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice when these mice were crossed with Igh7^−/− mice (Fig. 2 c). The total number of cells found in the spleen and LNs was almost completely normalized by the deficiency in IgE (Fig. 2 d). Analysis of the plasma (CD138⁺) cells in the spleen and LNs revealed a marked decrease in their numbers in the FcγRIIB^−/− Igh7^−/− and FcγRIIB^−/−.Yaa Igh7^−/− mice relative to their IgE-producing counterparts, and was similar to the plasma cell numbers in WT mice (Fig. 2 e). Moreover, although no significant difference was seen for CD19⁺B220⁺ B cells in the spleen of FcγRIIB^−/− and FcγRIIB^−/−.Yaa relative to their IgE-deficient counterparts, a significant decrease in their numbers was detected in the LNs of these mice and there was a strong trend for a reduction in these cells in both the spleen and LNs of FcγRIIB^−/− Igh7^−/− mice (Fig. 2 f). Collectively, the findings demonstrate the influence of IgE on the B cell compartment and the production of Ig in these lupus-prone mice.

The decrease in numbers of B and plasma cells in the spleen and LNs of IgE-deficient lupus-prone mice suggested the possibility that T cell responses might also be affected in these mice. To explore this, we first determined the number of CD4⁺ and CD8⁺ T cells in the spleen and LNs of these mice. The absolute numbers of CD4⁺ or CD8⁺ T cells in the spleen were increased in FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice (unpublished data). A significant increase in CD4⁺/CD8⁺ T cell ratio was observed in the spleen of FcγRIIB^−/−.Yaa mice with a trend toward an increased ratio in the spleen of FcγRIIB^−/− mice (Fig. 2 g). In both cases, IgE deficiency caused a decrease in this ratio to WT levels. No significant differences were observed in the LNs. Activation of both CD4⁺ and CD8⁺ cells, as determined by CD69 expression, was altered with a significant or strong trend for decreased activation of CD4⁺ and CD8⁺ cells, respectively, in the spleen of the IgE-deficient mice (unpublished data). In LNs, there also appeared to be a trend toward decreased activation of CD4⁺ and CD8⁺ cells with IgE deficiency that was more apparent for FcγRIIB^−/− relative to FcγRIIB^−/−.Yaa mice (unpublished data). These findings demonstrate that, in the absence of IgE, T cell responses appear to normalize.

Given that the loss of IgE resulted in reduced numbers of T and B cells and also caused some reduction in the activation of the former, we investigated whether this deficiency also affected other inflammatory cell types. The α-subunit of the high-affinity receptor for IgE (FcεRI) seems to be expressed on other cell types, including dendritic cells, eosinophils, monocytes, and neutrophils (Fig. 3 a; Gould and Sutton, 2008; Hammad et al., 2010; Porcherie et al., 2011). Ly6c^hi (inflammatory monocytes) constitutively expressed the highest levels of FcεRI (Fig. 3 a). Changes in FcεRI expression were minor-to-modest with increased expression in dendritic cells, eosinophils, and neutrophils relative to WT numbers. Dendritic cells, eosinophils, and neutrophils showed some association between the levels of FcεRI expression and IgE sufficiency (Fig. 3 a). Analysis of dendritic cell numbers in the spleen and LNs of FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice showed that their numbers increased in the secondary lymphoid organs (SLO’s) for both strains but was completely normalized by IgE deficiency (Fig. 3 b). Interestingly, although dendritic cell numbers were normalized in FcγRIIB^−/− Igh7^−/− and FcγRIIB^−/−.Yaa Igh7^−/− mice, activation of these cells as determined by CD86 expression appeared to be unaltered (Fig. 3 c).

Two populations of monocytes (Ly6c^lo-inter and Ly6c^hi, respectively) were identified from an F4/80⁺ Ly6G⁻ gated population. The numbers of resident (Ly6c^lo-inter) monocytes in the spleen and LNs was markedly increased in both FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice (Fig. 3 d). The number of resident monocytes returned to normal levels (those of WT mice) in FcγRIIB^−/− Igh7^−/− and FcγRIIB^−/−.Yaa Igh7^−/− mice (Fig. 3 d). Similarly, whereas the levels of inflammatory monocytes (Ly6c^hi) trended toward an increase in FcγRIIB^−/− mice and was significantly increased in FcγRIIB^−/−.Yaa mice, the apparent increases were normalized (to WT levels) in IgE deficiency (Fig. 3 e). This was also the case when neutrophils were analyzed (Fig. 3 f). Analysis of the numbers of eosinophils in the SLO’s revealed their presence only in the spleen of these mice and showed a strong trend toward increased numbers in both FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice that was reversed by IgE deficiency (Fig. 3 g). Thus, the findings demonstrate that IgE antibodies contribute to the recruitment of inflammatory cells in SLOs and suggests that IgE is involved in the inflammation seen in lupus-prone mice.

We have previously demonstrated that basophils can participate in the amplification of the inflammatory response in lyn^−/− mice by binding IgE autoantibodies (Charles et al., 2010). In this model, IgE deficiency caused a marked reduction in the numbers of basophils seen in SLO’s with improvement in disease phenotype. In FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice, we did not detect sufficient numbers of basophils in LNs to study the impact of IgE deficiency in this organ. Interestingly, however, FcεRI expression on splenic and blood basophils from FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice was decreased and this was markedly reversed by IgE deficiency (Fig. 4 a), suggesting that the engagement of this receptor on basophils by IgE causes a down-regulation of its cell surface expression as described for human basophils (MacGlashan et al., 1997). We further explored if the numbers of basophils in the spleen might be altered with disease development in FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice. As shown in Fig. 4 b, splenic basophil numbers appeared to be unaltered in FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice regardless of whether these mice produced IgE or not. However, a direct assessment of their activation status by measuring the levels of CD200R (a molecule that is highly expressed upon basophil activation; Torrero et al., 2009) demonstrated a marked reduction in its expression in basophils from FcγRIIB^−/−.Yaa Igh7^−/− mice (Fig. 4 c). Changes in splenic basophil expression of CD200R in FcγRIIB^−/− Igh7^−/− were likely not caused by the decreased numbers of basophils (Fig. 4 b) in the spleen of these mice. Nonetheless, basophil activation in the aggressive FcγRIIB^−/−.Yaa disease model is significantly dependent on the presence of IgE.

Previous work demonstrated that in human SLE the presence of autoreactive IgE was tightly linked to enhanced disease activity, increased serological markers, and lupus nephritis (Charles et al., 2010; Dema et al., 2014). Herein, we grouped human SLE subjects on whether they were positive or negative for autoreactive IgE to determine if significant differences could be observed in the association of these individual populations with markers of disease activity. As shown in Table 1, positivity for autoreactive IgE was associated with a significant increase in the proportion of individuals with active disease, increased serological markers, and lupus nephritis. Increased basophil activation in human SLE subjects (as demonstrated by enhanced CD203c expression; Bühring et al., 2004) was also previously demonstrated (Charles et al., 2010), and herein we found that increased CD203c expression is associated with mild or active SLE (Fig. 5 a). CD203c expression was also found to be significantly elevated above healthy controls in human SLE subjects that were positive for autoreactive IgE (Fig. 5 b). Moreover, in this group of subjects, CD203c expression was also significantly increased above healthy controls when individuals had active disease (Fig. 5 c). Collectively, these results implicate the presence of autoreactive IgE as a key factor in basophil activation in human SLE subjects. This is supported by our previous finding that basophils from human SLE subjects have enhanced phosphorylation of the FcεRγ (Suzuki et al., 2013), suggesting that the activation of human basophils in lupus is mediated through engagement of FcεRI by autoreactive IgE complexes.

Here, we provide evidence of a role for IgE in nonallergic inflammation and demonstrate that the presence of IgE facilitates the onset of disease development in lupus-prone mice and is tightly linked to basophil activation and active human SLE. This view is strongly supported by our findings in the highly accelerated and aggressive lupuslike disease model of FcγRIIB^−/−.Yaa mice, where we found that IgE deficiency delays the onset of disease, prolonging their life span. However, FcγRIIB^−/−.Yaa Igh7^−/− mice eventually develop a disease with many of the same phenotypic features of FcγRIIB^−/−.Yaa mice (unpublished data). Thus, although disease can develop in the absence of IgE, IgE appears to play a key role in amplifying the loss of tolerance in this model. This is supported by our human studies where individuals with increased autoreactive IgE were disproportionally linked to active disease.

Unexpectedly, our findings reveal an important and unrecognized role for IgE in the inflammatory response by promoting immune cell recruitment to SLO’s. The presence of FcεRI on various immune cell types during inflammation is likely to underlie this immunoregulatory role for IgE in the inflammatory response but we cannot exclude a role for CD23 from our studies to date. Basophils, which highly express FcεRI, are known to cooperate with DC’s in mounting an inflammatory response (Tang et al., 2010). Thus, as shown herein, in the absence of autoreactive IgE, basophil activation, and immune cell recruitment is dampened. Ultimately, this temporal defect is overcome as FcγRIIB^−/−.Yaa Igh7^−/− can eventually develop disease (unpublished data). Nonetheless, the findings argue that IgE provides a microenvironment that facilitates a rapid and potent inflammatory response by promoting immune cell activation and function in SLO’s. This reveals a previously unrecognized immunoregulatory role for IgE.

All experiments were performed following the animal study protocol A010-04-03 and A012-11-01 approved by the Laboratory Animal Care and Use section in accordance of National Institutes of Health (NIH) and National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS). C57BL/6 FcγRIIB^−/− and FcγRIIB^−/−.Yaa mice were previously described (Bolland and Ravetch, 2000; Bolland et al., 2002) and the IgE-deficient (Igh7^−/−; IgE heavy chain locus) were generously provided by H. Oettgen (Harvard University, Cambridge, MA; Oettgen et al., 1994) and backcrossed to C57BL/6 (N10). Double-deficient and the respective WT mice were generated after crossing FcγRIIB^−/− and FcγRIIB^−/−.Yaa with Igh7^−/− mice for more than four generations, and all genotypes used came from the same litters. Unless otherwise stated, mice of the different genotypes (FcγRIIB^−/−.Yaa, FcγRIIB^−/−.Yaa/Igh7^−/−, WT female only) were used at 16 wk of age, whereas 28-wk-old female mice were used for FcγRIIB^−/−, FcγRIIB^−/−/Igh7^−/−, and corresponding WT genotypes.

Blood samples were obtained from previously described cohorts of healthy controls and SLE subjects (Dema et al., 2014) as approved by the Institutional Review Board of NIAMS and by the Comité Régional de Protection des Personnes (CRPP, Paris, France). A written consent was obtained from all healthy donors and SLE patients. Parameters for human basophil identification, FcεRI, and CD203c detection were previously described (Charles et al., 2010). All sample processing was as previously described and the analysis was based on autoreactive IgE to the four most common lupus antigens (dsDNA, Sm, SSA/Ro, and SSB/La; Dema et al., 2014).

The relative levels of dsDNA-specific IgE and IgG were measured as previously described (Charles et al., 2010). For measurement of anti-histone and RNP IgGs, MaxiSorp plates (Thermo Fisher Scientific) were coated with 2 µg of whole histones or RNP (ImmunoVision) in PBS overnight at 4°C. After a washing step, the plates were blocked with PBS containing 10% FBS (HyClone) and 0.05% Tween 20 (Sigma-Aldrich). Plates were incubated with serum samples diluted in blocking buffer. Horseradish peroxidase-conjugated antibody to mouse IgG (goat anti–mouse IgG-Fc fragment; Bethyl Laboratories) was used for detection. Colorimetric detection was with tetramethylbenzidine substrate (Invitrogen) and the optical density was measured at 450 nm. Total Ig levels (IgG1, IgG2b, and IgG3) were determined by ELISA kit following the manufacturer’s instruction (Alpha Diagnostic International). Nuclear staining was performed to determine antinuclear Ig (ANA⁺), after incubation of mouse sera on fixed HepG2 cells.

To measure C3 and IgG deposits in the kidney, frozen sections (American Histolabs) were fixed with 4% paraformaldehyde solution and blocked with PBS supplemented with 7.5% Blokhen II (Aves Labs) for 1 h. Sections were incubated with FITC-conjugated anti–mouse complement C3 (Cedarlane) and FITC-conjugated affiniPure F(ab′)₂ fragment goat anti–mouse IgG, Fcγ-specific (Jackson ImmunoResearch Laboratories) or the respective FITC-conjugated isotype controls (FITC-rat IgG2a isotypic control; Cedarlane; FITC-goat IgG isotype control; Abcam) overnight at 4°C. The sections were washed with PBS containing 0.1% Tween 20 (Sigma-Aldrich). After washing the tissue, the slides were embedded in mounting media (Vector Laboratories, Inc.). FITC-conjugated antibodies were excited with an argon ion laser at 488 nm and fluorescence emission of >505 nm was measured with a confocal laser fluorescence microscope (LSM-780; Carl Zeiss, Inc.). Image processing was performed using ImageJ (NIH) or Photoshop (Adobe Systems) software. In these experiments, one of the two kidneys was embedded in 10% neutral buffered formalin (Sigma) and then sectioned and H&E stained by American Histolabs. Nephritis assessment and scoring from 0 to 4 (severe lesions) was established in a blinded manner by a pathologist assessing glomeruli and tubular injury. Pictures were taken on the Keyence BZ-9000E fluorescence microscope (BIOREVO).

Kidney function was assessed by measuring albumin (Bethyl Laboratories) and creatinine (R&D Systems) ratio (ACR; microgram of albumin per microgram of creatinine) in urine of the different strains.

Blood, spleen, and LNs (cervical, inguinal, axial, and brachial) were obtained from the mice. Spleen and LNs were homogenized and filtered through 70-µm cell strainer (BD Falcon) for single-cell suspensions. Red blood cells (for blood and spleen) were lysed with ammonium-chloride-potassium lysing buffer (Quality Biological, Inc.) for 5 min on ice, three times. Specific cell surface markers were detected by incubation for 25 min at 4°C with fluorescently conjugated antibody in FACS buffer (PBS, 0.1% BSA, 0.05% NaN₃) after blocking low affinity Fcγ receptors with CD16/32 antibody (BioXcell). When biotinylated antibody was used a subsequent incubation for 15 min with PerCP/Cy5.5-conjugated Streptavidin (eBioscience) was performed. Cells were stained with the Aqua dye fluorescence live/dead fixable stain kit (Invitrogen), following the manufacturer’s instructions. FMO (Fluorescence minus one) strategy was used to determine if the detected fluorescence indicated positivity for a cell surface marker or not. Flow cytometry acquisition was on a FACSCanto II (BD) and analysis was done using FlowJo software (Tree Star). Antibodies used from eBioscience were as follows: PE-anti–mouse TCRβ (clone: H57-597), FITC-anti–mouse CD4 (clone: GK1.5), APC-anti–mouse CD69 (clone: H1.2F3), FITC-anti–mouse CD23 (clone B3B4), Biotin-anti–mouse CD123 (clone: 5B11), and APCeFluor780-anti–mouse MHC II (I-A/I-E; clone: M5/114.15.2). Antibodies used from BioLegend were as follows: Pacific Blue-anti–mouse CD19 (Clone: 6D5), APC/Cy7 anti–mouse B220 (clone: RA3-6B2), PE/Cy7-anti–mouse CD21/CD35 (CR2/CR1; clone: 7E9), PE-anti–mouse CD138 (Syndecan-1; clone:281-2), PerCP/Cy5.5-anti–mouse IgM (clone: RMM-1), APC-anti–mouse CD49b (pan-NK cells; clone: DX5), Pacific blue-anti–mouse CD117 (c-kit; clone: 2B8), PE-anti–mouse FcεRIα (clone: MAR-1), FITC-anti–mouse CD200R (OX2R; clone: OX-110), PECy7-anti–mouse CD62L (clone: MEL-14), PECy7-anti–mouse Ly6G (clone: 1A8), PerCP/Cy5.5-anti–mouse CD11c (clone: N418), Pacific Blue-anti–mouse CD11b (clone: M1/70), and Alexa Fluor 640-anti–mouse F4/80 (clone: CI:A3-1). Antibodies used from BD were as follows: PerCP/Cy5.5-anti–mouse CD8a (clone: 53–6.7), FITC-anti–mouse Ly6C (clone: AL-21), FITC-anti–mouse CD86 (clone: GL1), and PE-anti–mouse CD80 (B7-1; clone: 16-10A1).

One-way ANOVA with Tukey’s correction test or Kruskal-Wallis with Dunn’s correction test was applied when more than two groups were compared. Mean and SEM are shown on the graphs. When a different statistical analysis was performed it is indicated in the figure legend. All analysis was performed using GraphPad Prism 6.0.

Gating strategy to define mouse myeloid cell types is shown in Fig. S1.

This research was supported by the Intramural Research Program of NIAMS and NIAID, NIH, by the French Institut National de la Santé Et de la Recherche Médicale (INSERM), by the Assistance Publique – Hôpitaux de Paris (AP-HP), and by grants from the Mairie de Paris (Emergences 2010), ANR JCJC SVSE1 2011 BASILE and the French Kidney Foundation (Fondation du Rein) to N. Charles. N. Charles and B. Dema are supported by the Laboratoire d’excellence INFLAMEX. We acknowledge the support of the Laboratory Animal Care and Use, Flow Cytometry, and Light Imaging Sections of the Office of Science and Technology, NIAMS.

The authors have no conflicting financial interests.