An autoimmune disease variant of IgG1 modulates B cell activation and differentiation

See allHide authors and affiliations

Science  09 Nov 2018:
Vol. 362, Issue 6415, pp. 700-705
DOI: 10.1126/science.aap9310

An IgG1 SNP enhances autoimmunity

One common feature of autoimmune diseases like systemic lupus erythematosus (SLE) is the presence of high titers of self-reactive antibodies. These result in immune complexes, inflammation, and tissue pathology. Consequently, the checkpoints that normally keep immunoglobulin G (IgG)–positive autoreactive B cells in check are of intense interest. Chen et al. report the presence of a common IgG1 single-nucleotide polymorphism (SNP) in East Asian populations (hIgG1-G396R). This SNP was enriched in SLE patients and associated with increased disease severity. Humans with this SNP, as well as knockin mice, showed enhanced plasma cell accumulation and antibody production. This SNP enhanced IgG1 immunoglobulin tail tyrosine motif phosphorylation, triggering longer adaptor protein Grb2 dwell times in immunological synapses and hyper–Grb2–Bruton's tyrosine kinase signaling after antigen binding.

Science, this issue p. 700


The maintenance of autoreactive B cells in a quiescent state is crucial for preventing autoimmunity. Here we identify a variant of human immunoglobulin G1 (IgG1) with a Gly396→Arg substitution (hIgG1-G396R), which positively correlates with systemic lupus erythematosus. In induced lupus models, murine homolog Gly390→Arg (G390R) knockin mice generate excessive numbers of plasma cells, leading to a burst of broad-spectrum autoantibodies. This enhanced production of antibodies is also observed in hapten-immunized G390R mice, as well as in influenza-vaccinated human G396R homozygous carriers. This variant potentiates the phosphorylation of the IgG1 immunoglobulin tail tyrosine (ITT) motif. This, in turn, alters the availability of phospho-ITT to trigger longer adaptor protein Grb2 dwell times in immunological synapses, leading to hyper–Grb2–Bruton’s tyrosine kinase (Btk) signaling upon antigen binding. Thus, the hIgG1-G396R variant is important for both lupus pathogenesis and antibody responses after vaccination.

Autoimmune diseases such as systemic lupus erythematosus (SLE) are characterized by the presence of large numbers of self-reactive antibodies that induce deposition of immune complexes (ICs), leading to inflammation and tissue damage (1). Autoreactivity is pervasive in the antibody repertoire of human B cells across different developmental stages (2). It is especially enriched in the peripheral immunoglobulin G–positive (IgG+) memory B cell pool but is efficiently diminished in the plasma cell compartment in healthy individuals (24). However, these checkpoints fail in patients with autoimmune diseases. It remains unclear how autoreactive IgG+ B cells are maintained in a quiescent state under physiological immune homeostasis and how these checkpoints are broken in pathological conditions. IgG–B cell receptor (IgG-BCR) potently enhances memory IgG antibody responses via the evolutionarily conserved cytoplasmic tail of membrane-bound IgG (mIgG-tail) (510). The mIgG-tail amplifies BCR signaling via its phospho–immunoglobulin tail tyrosine (ITT) motif, which recruits the adaptor protein Grb2 to enhance Ca2+ mobilization, synergistically with Bruton’s tyrosine kinase (Btk) and phospholipase C–γ2 (PLC-γ2) (5, 11).

Here we identified a single-nucleotide polymorphism (SNP) rs117518546, which results in a glycine-to-arginine substitution at codon 396 in human IgG1 (hIgG1-G396R) (fig. S1A and table S1). This SNP was common in East Asian populations (fig. S1B) and was significantly correlated with susceptibility to SLE (tables S2 and S3). The G396R variant frequency was substantially enriched in SLE patients compared with that of criteria-matched controls in three independent cohorts from multiclinical centers in China (1786 healthy controls versus 1838 SLE patients in total, Pearson’s chi-square test and binary logistic regression analysis, P = 6.0 × 10−5). Furthermore, the G396R variant was associated with a more severe disease phenotype, including earlier onset, multiple organ involvement, and higher SLE disease activity; specifically aggravated autoantibody production; and inflammation (Fig. 1A and table S4). The variant drove an autoantibody subclass profile shift toward IgG1-isotype predominance in G396R patients (Fig. 1B). Thus, hIgG1-G396R is a risk locus for SLE.

Fig. 1 An SLE-correlated SNP variant increases autoantibody production.

(A) Genotypic correlation analysis of the human IgG1-G396R variant with clinical manifestations in SLE patients (279 controls and 251 SLE patients, as detailed in table S4). The percentages indicate the portion of G396R homozygotes. Hetero, heterozygous; SSA, Sjögren’s syndrome–related antigen A; SSB, Sjögren’s syndrome–related antigen B; r-RNP, ribonucleoproteins. (B) Ratio of IgG subclass for ANA in WT (n = 22), Hetero (n = 19), and G396R (n = 13) SLE patients. One-way analysis of variance (ANOVA), IgG1/IgG2 (P = 0.0017), IgG1/IgG3 (P = 0.0552), and IgG2/IgG3 (P = 0.9268). (C) Anti-dsDNA antibodies detected in serum from bm12-induced WT (n = 7) and G390R (n = 5) mice 3 weeks after induction. OD490, optical density at a wavelength of 490 nm. (D) Immunofluorescence of IgG1+ ANA with HEp-2 reactivity in week 3 serum from bm12-induced mice (WT, n = 6; G390R, n = 6). Mean fluorescence intensity (MFI) of IgG1 and IgG2b subclass ANA was quantified. DAPI, 4′,6-diamidino-2-phenylindole. (E) Autoantigen microarray was used to detect autoantibodies in serum as in (C). (F) Immunofluorescence of deposited IgG1 ICs in mice glomeruli at week 6 after induction. Statistical analysis of MFI of IgG1 and IgG2b ICs in glomeruli. The scale bar represents 20 μm in (D) and (F). Unpaired two-tailed Student’s t tests, (B) to (D) and (F). NS, not significant. Red bars indicate means. At least three sections were analyzed for each sample in (F). Data are representative of at least two independent experiments in (B) to (D) and (F).

We generated knockin mice harboring the murine homolog mIgG1-G390R (denoted as Ighg1T/T or G390R mice) (figs. S1C and S2, A and B). Under normal conditions, there were no significant differences in terms of natural antibodies or other evident phenotypes between wild-type (WT) and G390R mice (fig. S2C). In the bm12 splenocyte–inducible lupus model (12), the levels of IgG1 subclass anti–double-stranded DNA (dsDNA), anti–Smith D (SmD), and antinuclear antibodies (ANA) were notably increased in G390R mice (Fig. 1, C and D, and fig. S3A). Autoantigen microarrays further confirmed a clear increase in IgG1 but not IgM and IgG2b autoantibody production in G390R mice (Fig. 1E and fig. S3B). Moreover, G390R mice showed enlarged glomeruli and substantial IgG1 deposits containing ICs along with moderate IgG2b deposition (Fig. 1F). Similar results were also observed in a second lupus model by using apoptotic thymocytes (fig. S3, C to E). In aged mice, the G390R variant facilitated autonomous IgG1 autoantibody production (fig. S3F). Thus, the G390R variant promotes autoantibody production during autoimmune disease progression.

Six weeks after the induction of autoimmunity with bm12 splenocytes, there was a fourfold increase in the number of IgG1+ germinal center (GC) B cells in G390R mice compared with that of WT mice (Fig. 2A and fig. S4A). IgG1+ memory B cells only mildly increased in G390R mice (Fig. 2B). Notably, there was an almost sixfold increase in IgG1+ plasma cells in both the spleens and bone marrow of G390R mice (Fig. 2, C to E, and fig. S4A). This phenomenon was also seen in models of apoptotic thymocyte-induced lupus and aging (fig. S4, B and C). In a competitive model, in which WT or G390R B cells were adoptively transferred into B cell–deficient hosts (fig. S4D), more IgG1+ plasma cells were differentiated in the G390R group than in the WT group relative to the internal control. The competition index for the G390R group was 50% higher than that for the WT group (fig. S4E). Thus, increased B cell differentiation upon antigen stimulation, especially plasma cell generation, may be responsible for excessive autoantibody production in G390R mice.

Fig. 2 The IgG1-G390R variant promotes plasma cell accumulation in induced autoimmune models.

(A to D) Flow cytometric analyses of IgG1+ GC B cells, memory B (Mem B) cells, spleen (SP), and bone marrow (BM) plasma cells (PC) in WT (n = 6) and G390R (n = 7) mice at week 6 after bm12 splenocyte induction. GC B cells (GL7+, Fas+) were pregated on B220+ cells. Percentage (means ± SEM) of B cell subsets and corresponding comparison of cell numbers (right) are indicated. (E) The relative percentage (means ± SEM) of IgG1+ plasma cells and memory B cells in fate-selected cells of WT in comparison to those of G390R mice. (F) Frequency (means ± SEM) of Igγ1 CDR3 with different numbers of positively charged amino acids in IgG1+ bone marrow plasma cells (WT CDR3, n = 64; G390R CDR3, n = 66) from bm12-induced WT (n = 3) and G390R mice (n = 3). Unpaired two-tailed Student’s t tests, (A) to (E). Two-way ANOVA, (F). Bars indicate means. Representative data are from at least two independent experiments in (A) to (F).

Positively charged amino acids in IgH complementarity-determining region 3 (CDR3) are associated with antibody autoreactivity (13), and indeed, these types of amino acids were enriched in IgG1+ plasma cells from G390R mice compared with those from WT mice (Fig. 2F). Moreover, the length of the Igγ1 CDR3, which potentially predicts autoreactivity enrichments (2), was prominently increased in the G390R population (fig. S4F). We also observed reduced Ig variable (V) gene VH1 and joining (J) gene JH1 usage, but increased VH5, VH14, and JH4 usage in G390R IgG1+ plasma cells (fig. S4G).

The enhanced antibody production by the hIgG1-G396R variant may also have effects on physiological humoral responses to non-self antigens. Thus, we conducted influenza vaccination experiments in healthy human WT controls and G396R homozygous carriers (Fig. 3A). The G396R variant significantly potentiated the generation of influenza virus–specific IgG1 (Fig. 3B) and only showed a relative minor effect on IgG2 production. We also confirmed these effects in 4-hydroxy-3-nitrophenylacetyl (NP) eight-keyhole limpet hemocyanin (KLH)–immunized mice (Fig. 3C). NP-specific IgG1 antibody responses were almost twofold higher in G390R mice and threefold higher during antigen recall responses (Fig. 3D). Affinity maturation did not appear to be significantly affected (fig. S4H). The increase in antibody production was in line with the enhanced numbers of NP-specific IgG1+ GC B and plasma cells in G390R mice (Fig. 3, E and F). Thus, this variant enhances antibody production in both autoimmune disorders and in physiological humoral responses upon vaccination.

Fig. 3 The IgG1-G390R variant facilitates IgG1 antibody production in physiological humoral responses.

(A) Schematic diagram of the human influenza vaccination study. WT and G396R homozygous healthy volunteers were influenza vaccinated on day 1. Pre- and postimmune sera were collected on day 0 and day 20. (B) Influenza-specific IgG1 and IgG2 antibody responses in vaccinated WT (n = 6) and G396R (n = 3) healthy volunteers. Fold change of antibody titers on day 20 to day 0 was quantified. (C) Diagrammatic representation of the NP8-KLH immunization in mice. GC B, memory B, and plasma cells were analyzed on days 7 (GC B) and 14 (memory B and plasma cells). (D) NP-specific antibody titers in WT (n = 7 or 3) and G390R (n = 5) mice on days 14 (left) and 42 (right) after NP8-KLH immunization. (E) Flow cytometric analyses of NP-specific IgG1+ GC (pregated on B220+, GL7+, CD38) on day 7, memory B cells (pregated on B220+, IgG1+, NP+), spleen and bone marrow plasma cells (pregated on B220lo, CD138+) on day 14 in NP8-KLH immunized WT (n = 4 or 5) and G390R (n = 4) mice. Percentage (means ± SEM) and B cell–subset numbers were compared. (F) Spleen and bone marrow plasma cell numbers in NP recall responses (WT, n = 3; G390R, n = 3). Unpaired two-tailed Student’s t tests, (C) to (F). Red bars indicate means. NS, not significant. Data in (C) to (F) are representative of at least two independent experiments.

Because IgG1+ light-zone GC B cells in bm12-induced G390R mice up-regulated plasma cell fate favoring transcription profiles (Fig. 4A), we then investigated the membrane-proximal signaling upon B cell activation (8, 14). BCR microcluster formation, the synaptic recruitment of Grb2 and Btk, calcium mobilization, and the phosphorylation of downstream signaling molecules [Erk, S6, and nuclear factor–κB (NF-κB)] were significantly enhanced, indicating heightened signaling mediated by the variant (Fig. 4, B and C, and fig. S5, A to H). This effect was also observed in IgG1+ primary B cells from the peripheral blood of G396R homozygotes compared with those from WT SLE patients (Fig. 4D and fig. S5, I and J).

Fig. 4 The IgG1-G390R variant induces excessive IgG-BCR signaling by potentiating phospho-ITT–Grb2 signaling module.

(A) Transcription profile of Bcl6, Irf4, Blimp1, c-myb, and Pax5 in IgG1+ light-zone GC B cells (B220+, GL-7+, Fas+, CD86hi, CXCR4lo) from WT (n = 3) and G390R (n = 3) mice upon immunization. (B) Total fluorescence intensity (FI) of synaptic accumulated BCRs were compared in IgG1+ class-switched WT and G390R primary B cells with different concentrations of antigen stimulation. (C) Dynamics of Grb2 recruitment to the immunological synapse in WT and G390R class-switched IgG1+ primary B cells (n > 13). (D) Synaptic IgG1-BCR accumulation and Grb2 recruitment in IgG1+ B cells from the peripheral blood of WT (n = 7) and G396R homozygous (n = 5) SLE patients after surrogate antigen stimulation. (E) Heat map indicates ratio dynamics of mCerulean to cpV (FRET channel) fluorescence intensity in cells expressing FRET-based phospho-ITT activation biosensor with WT or G390R ITT motifs after antigen stimulation. Ratio dynamics were quantified, and ratios were normalized to the initial frame upon activation. (F) Trajectories of mEos3.1-Grb2 (colored) in immunological synapse (gray). Dwell time of mEos3.1-Grb2 in activated IgG1+ class-switched WT or G390R primary B cells (n > 10) are calculated and compared, with resting WT cells as control. (G) In vitro phosphorylation assay with purified glutathione S-transferase (GST)–fused mIgG-tail protein to compare the phosphorylation of WT or G390R ITT motif by active Lyn kinase. p, phosphorylated. (H) Predicted ITT motif peptide binding modes with the active Lyn kinase domain. A snapshot of G390R ITT motif:Lyn kinase complex structure from molecular dynamics simulations and molecular mechanics–generalized Born surface area (MM-GBSA) binding free energy (ΔGbind) are shown. (I) Probability distributions of the distance between the hydroxyl group of catalytic residue Tyr385 and the γ-phosphate group of ATP in 100-ns simulations with WT or G390R ITT motif peptide docking to the active Lyn kinase domain in 50 to 100 ns of the simulations. The scale bars represent 1.6 μm in (C) to (F). Unpaired two-tailed Student’s t tests in (B) to (D) and (F). Two-way ANOVA in (E). Red bars indicate means. NS, not significant. Data are representative of three independent experiments in (A) to (G).

In a fluorescence resonance energy transfer (FRET)–based phospho-ITT activation reporter (fig. S6, A to C), G390R variants showed significantly higher FRET ratio changes compared with WT upon antigen stimulation (Fig. 4E). Thus, the G390R variant promotes the activation of the phospho-ITT–Grb2 signaling module. Additional single-molecule imaging experiments revealed a more confined Grb2 motion in the immunological synapse of the G390R variant compared with that of WT primary B cells (Fig. 4F). Thus, two potential models were proposed to explain the excessive activation of phospho-ITT–Grb2 signaling module: Increased Grb2 binding to the phospho-ITT motif and/or increased ITT-motif phosphorylation. Isothermal titration calorimetry assays demonstrated similar binding between the Grb2 protein and either the WT or G390R phospho-ITT motif peptide (fig. S7A). In the second model, both Syk and Lyn tyrosine kinases are involved in the phosphorylation of the ITT motif (11). G390R ITT phosphorylation by Syk was only mildly higher (fig. S7, B and C). Substantially enhanced phosphorylation of the G390R ITT motif by Lyn was observed in vitro (Fig. 4G and fig. S7D) and in vivo (fig. S7E).

Finally, to identify the structural basis for excessive G390R ITT motif phosphorylation by Lyn, we carried out molecular dynamics simulations of the Lyn kinase domain bound to either the WT or the G390R ITT motif. We constructed complex models of substrate peptide and kinase domain based on the structure of the active human Lyn kinase domain (Protein Data Bank 3A4O) (15). The antiparallel β sheet formed by the substrate peptide and the activation loop of kinase domain persisted for the length of each trajectory. Additionally, the Asp-Phe-Gly (DFG) motif and αC helix of the Lyn kinase domain were maintained in an active conformation (Fig. 4H and fig. S7, F and G). Notably, the Arg390 residue in the G390R ITT motif formed an additional hydrogen bond with the backbone carbonyl of Asn290 in Lyn kinase domain, which was not available for the Gly390 residue in the WT ITT motif (Fig. 4H). Furthermore, the distance between the hydroxyl group of catalytic residue Tyr385 and γ-phosphate group of adenosine triphosphate (ATP) was significantly shorter for the G390R variant than for WT (Fig. 4I and fig. S7H), suggesting more stable contacts and potentially more effective phosphotransfer for the G390R variant. Indeed, the binding free energy of the G390R variant was consistently lower than that of the WT ITT motif (Fig. 4H). Together, these computational results suggest the higher binding ability of the G390R variant to active Lyn kinase.

The G390R variant potentiates ITT-motif phosphorylation and alters the availability of the phospho-ITT motif, thereby triggering a significantly longer dwell time of Grb2 in the immunological synapse. This, in turn, changes the synaptic Grb2 recruitment from a “recruit-and-escape” model to a “recruit-and-confine” model, in which the initially recruited but subsequently escaped Grb2 can be recaptured by the proximal phospho-ITT motifs within the synapse (fig. S8). This results in excessive G390R variant IgG1+ B cell activation and differentiation to plasma cells producing increased IgG1 antibodies in the context of both autoimmune disorders and physiological humoral responses upon infection or vaccination. This enhancement of G390R variant IgG1+ B cells may have some bona fide effects on the production of other IgG subclass antibodies by potential mechanisms: (i) Enhanced IgG1+ B cell activation may promote the function of T follicular helper cells (16) to potentiate the activation and differentiation of other IgG subclass B cells; (ii) increased IgG1+ antibodies may drive the activation and proliferation of dendritic cells, which then enhance general B cell activation and differentiation (17); (iii) IgG1+ B cells may undergo class switching again to other IgG subclasses (18); (iv) IgG1+ B cells may produce inflammatory cytokines such as interleukin-12 (IL-12) and tumor necrosis factor–α (TNF-α) to regulate the differentiation of other IgG-subclass plasma cells (19); and (v) excessive IgG1 antibodies may promote antigen presentation to activate B cells producing other IgG-subclass antibodies (20). Thus, we contend that IgG-BCR not only promotes enhanced memory responses but also leads to the development and relapse of autoimmune diseases when dysregulated.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

Tables S1 to S4

References (2127)

References and Notes

Acknowledgments: We thank S. K. Pierce, J. Lukszo, and K. Rajewsky for providing experimental materials. We thank Q. Z. Li and I. Raman in Microarray Core Facility, University of Texas Southwestern Medical Center, for support with the autoantigen microarray. We thank X. Wang, J. Wu, and Z. Wang for discussions. We thank P. Tolar, C. Wu, and D. Long for critical reading of this manuscript. Funding: This work is supported by funds from the National Natural Science Foundation of China (8182500030, 81730043, and 81621002 to W.L. and 31530020 to Z.L.), Ministry of Science and Technology of China (2014CB542500-03 to W.L. and 2014CB541901 to J.G.), Beijing Sci-Tech Program (Z171100000417007 to Z.L.), Beijing Nova Program (Z171100001117025 to X.S.), and Sanming Project of Medicine in Shenzhen (SZSM201612009 to Q.W.). Author contributions: Conceptualization, X.C. and W.L.; funding acquisition, W.L., Z.L., X.S., J.G., and Q.W.; investigation, X.C., B.Y., L.H., S.C., L.X., C.Y., W.Y., X.Zha., J.G., J.H., X.Zho., H.Z., Q.W., and H.S.; methodology, X.C., X.S., and W.Y.; resources, W.Z., Y.S., Z.C., Z.S., Z.H., F.Zha., F.Zhe., H.C., and J.L.; supervision, W.L., Z.L., and L.L.; writing–original draft, X.C. and W.Y.; and writing–review and editing, W.L., Z.L., X.C., B.Y., X.S., L.L., Z.Y., H.Q., and C.X. All authors contributed to revising the manuscript. Competing interests: The authors declare no financial or commercial conflicts of interest. Data and materials availability: All data described in this paper are present either in the main text or in the supplementary materials.

Stay Connected to Science

Navigate This Article