Human DC-SIGN and CD23 do not interact with human IgG

Abstract

The precise mechanisms underlying anti-inflammatory effects of intravenous immunoglobulin (IVIg) therapies remain elusive. The sialylated IgG fraction within IVIg has been shown to be therapeutically more active in mouse models. Functionally, it has been suggested that IgG undergoes conformational changes upon Fc-sialylation which sterically impede binding to conventional FcγRs, but simultaneously allow binding to human DC-SIGN (SIGN-R1 in mice) and also CD23. These latter C-type lectins have been proposed responsible for the immunomodulatory effects in mouse models. However, there is conflicting evidence supporting direct interactions between sialylated human IgG and CD23/DC-SIGN. While cells expressing human CD23 and DC-SIGN in their native configuration bound their natural ligands IgE and ICAM-3, respectively, no IgG binding was observed, regardless of Fc-glycan sialylation in any context (with or without bisection and/or fucosylation) or presence of sialylated Fab-glycans. This was tested by both by FACS and a novel cellular Surface Plasmon Resonance imaging (cSPRi) approach allowing for monitoring low-affinity but high-avidity interactions. In summary, we find no evidence for human CD23 or DC-SIGN being bona fide receptors to human IgG, regardless of IgG Fc- or Fab-glycosylation status. However, these results do not exclude the possibility that either IgG glycosylation or C-type lectins affect IVIg therapies.

Figure 1

Generation of glyco-engineered human IgG1. (a) Schematic representation of glyco-engineered IgG1 antibodies generated in this study with their biantennary glycan on asparagine (Asn) 297 (white star) in the Fc CH2 domain and, for the Fab-glycoform, on Asn 29 and Asn 86 (orange stars) in the VH and VL domain, respectively. (b) Bar graph summarizing Fc-glycoform N-glycan profile of variable glycan end groups (fucose (+/−F), in red; bisecting GlcNAc (+/−B) in cyan; mono-/di-galactose in pale/dark yellow; mono-/di-sialic acid in pale/dark purple) determined by mass spectrometry as percentages (%) of total glycosylation. N-glycan structures are schematically represented; galactose (yellow circle); mannose (green circle); N-acetylglucosamine (blue square); bisecting N-acetylglucosamine (cyan square); fucose (red triangle); sialic acid (purple diamond). (c) Bar graph showing the level of sialic acid of Fab-glycosylated anti-TNP as determined by Sambucus nigra agglutinin (SNA)-lectin ELISA with high sialic acid Fab-glycoform (grey bar) to asialylated (unmodified) IgG1 without a Fab-glycan (white bar). Data are representative of three independent experiments (in triplicate) showing mean ± standard error of the mean (s.e.m.). Statistical analysis was performed by a Two-tailed paired t-test (**P < 0.01).

Figure 2

CD23 and DC-SIGN are functionally expressed on transfected HEK Freestyle cells, but do not bind human IgG1, regardless of sialylation status. (a–c) HEK Freestyle cells were transfected with human FcγRIIa (CD32, red), CD23 (blue), DC-SIGN (green), or an empty vector (Mock, pink) and tested for binding to anti-CD32, anti-CD23. or anti-DC-SIGN, or their natural ligands at 37 °C: asialylated IgG1, IgE, or ICAM-3, respectively. (d) Binding of different soluble IgG1 glycoforms (10 μg/ml) to HEK CD23, DC-SIGN, FcγRIIa and Mock transfected cells at 37 °C. Data are representative of three independent experiments (in duplicate) and shown as mean values of the geometric mean fluorescence intensity (gMFI ± s.e.m.) with the background signals subtracted. Significant differences compared to HEK Mock signals were determined by One-way ANOVA with Dunnett’s multiple comparisons test (*P < 0.05, **P < 0.01, ****P < 0.0001).

Figure 3

Cellular SPR imaging layout and detection of receptor expression. (a) Schematic layout of the experimental setup for the cellular SPR, with HSA-TNP antigens depicted as red circles, SPR-sensor as yellow block, receptor-expressing cells as yellow circles with blue receptors and increasing gradient flow with a dashed arrow. (b) Representative sensorgram of five cellular SPRi runs on HSA-TNP- (10 nM) and HSA- (100 nM) (negative control) coated spots. Numbers depicted in the sensorgram correspond to the schematic representations in A, which shows the processes on the HSA-TNP spots at different time points. 1 IgG1 glycoform (asialylated in this case) injection; 2 Anti-TNP glycoform binding to HSA-TNP spot; 3 Receptor-bearing cell injection; 4 and 5 Cell sedimentation phase and start gradient flow respectively. 4a, 5a show cell binding and 4b, 5b no binding scenario. Green, blue and red backgrounds in the sensorgram represent the anti-TNP glycoform immobilization, cell sedimentation and gradient flow (1–120 µl/s) phase, respectively. Each line represents the binding of a specific cell type (HEK FcγRIIa in red, HEK Mock in pink, HEK CD23 in blue, HEK DC-SIGN in green for the HSA-TNP spots) in resonance units (RU) to immobilized asialylated anti-TNP IgG1 through time, with buffer flow (in yellow) to determine the background signal. (c) Representative sensorgrams show the binding of HEK FcγRIIa, HEK CD23 and HEK DC-SIGN cells to spots coated with murine antibodies against CD32, CD23 and DC-SIGN, respectively. These spots functioned as cell flow controls to confirm the presence of receptors. Significant differences compared to control (HEK Mock (panel B) or isotype control spots (panel C)) were based on the RU values on the last time point of the flow phase, determined by One-way ANOVA with Tukey’s multiple comparisons test (**P < 0.01, ***P < 0.001, ****P < 0.0001). Data are representative of three independent experiments showing mean ± standard deviation (s.d.). Each line represents the average sensorgram, with s.d. as dotted lines, from at least three spots monitored in real time simultaneously.

Figure 4

Neither human CD23- nor DC-SIGN-expressing cells bind human IgG1-containing immune complexes, regardless of glycosylation status. Cell flow sensorgrams after spot-specific buffer flow (background) subtraction (ΔRU). (a–g) Sensorgrams show binding responses to the different HSA-TNP (10 nM)-immobilized IgG1 glycoforms (specific glycan depicted in inset) of HEK DC-SIGN and HEK CD23 compared to HEK Mock with HEK FcγRIIa as positive control. Line identities and sugar moieties are represented by different shapes as indicated in the figure legend. Data are representative of three independent experiments showing mean ± s.d. Each line represents the average sensorgram, with s.d. as dotted lines, from at least three spots monitored in real time simultaneously. Significant differences compared to HEK Mock were based on the RU values on the last time point of the flow phase, determined by One-way ANOVA with Tukey’s multiple comparisons test (***P < 0.001).