FcγRIIB on liver sinusoidal endothelial cells is essential for antibody-induced GPVI ectodomain shedding in mice

The platelet collagen receptor glycoprotein VI (GPVI) is a promising pharmacological target because its absence or functional inhibition provides protection from thrombosis and stroke in several mammalian model organisms^1-3  without causing major bleeding.⁴  Likewise, GPVI deficiency in humans resulting from genetic defects^5,6  or autoimmunity does not cause a bleeding diathesis.^7,8  GPVI can be efficiently depleted in circulating platelets in mice and humans by antibodies.⁹  The major route of GPVI immunodepletion is ectodomain shedding, which requires intact GPVI signaling and is accompanied by a transient drop in platelet count.¹⁰  Under conditions of defective GPVI signaling, the receptor is still downregulated in vivo, but this occurs considerably more slowly through internalization/intracellular degradation, in the absence of thrombocytopenia. Importantly, however, although different anti-GPVI antibodies efficiently deplete the receptor in vivo in mice, they do not cause GPVI downregulation in vitro.¹¹ 

The Fc portion of immunoglobulin G (IgG) interacts with Fcγ receptors (FcγRs) on different cells and thereby triggers immune effector functions in vivo. In mice, 3 activating (FcRγ-chain/immunoreceptor tyrosine-based activation motif–linked) FcγRs exist, namely FcγRI, FcγRIII, and FcγRIV, and 1 immunoreceptor tyrosine-based inhibitory motif –bearing inhibitory receptor, FcγRIIB.¹²  FcγRIIB, which binds IgG immune complexes, but not monomeric IgG, is coexpressed with the activating FcγRs in different immune cells, such that the balance in their expression determines the strength of FcγR activation in response to IgG-immune complexes. However, the majority of FcγRIIB is found in liver sinusoidal endothelial cells (LSECs), which lack activating FcγRs.¹³  LSEC-FcγRIIB is the critical scavenging receptor mediating capture and removal of viruses and soluble immune complexes.^13,14  However, a role of FcγRIIB in the processing of antibody-opsonized blood cells including platelets has not been reported.

Knockout mice for GPVI (Gp6^−/−) FcγRIIB (Fcgr2b^−/−) and FcγRIII (Fcgr3^−/−) were described previously.^15-17  Animal studies were approved by the district government of Lower Franconia.

Streptavidin-horseradish peroxidase (DAKO), clodronate (clodronateliposomes.com), and rabbit anti-rat-IgG-fluorescein isothiocyanate (Dianova) were purchased. JAQ1 (anti-GPVI) and 2.4G2 (anti-CD16/32) were produced and modified in our laboratory.

Experiments were performed as described previously.¹⁰ 

Circulating platelets were labeled with 0.3 µg/g body weight (BW) anti-GPIX^AF750 antibody IV. Four hours later, mice were anesthetized with medetomidine (Pfizer; 0.5 µg/g), midazolam (Roche Pharma AG; 5 µg/g BW), and fentanyl (Janssen-Cilag GmbH; 0.05 µg/g BW). Subsequently, JAQ1 (2 µg/g BW) or vehicle was injected IV and in vivo imaging was performed on an IVIS spectrum device (Perking Elmer) at 800 nm (emission filter).

Liver samples were embedded in Tissue-Tek (Sakura Finetek) and snap-frozen in liquid nitrogen. Unspecific binding was blocked with 3% BSA/PBS.

Immunohistochemistry stainings were performed as described previously.¹⁸ 

Immunofluorescence (IF) stainings were performed as follows. Samples were stained with the indicated fluorophore-conjugated antibodies, counterstained with 4′,6-diamidino-2-phenylindole and visualized using confocal microscopy (Leica TCS-SP5). Images were processed in a blinded manner by thresholding, generation of binaries, using the “particle analyzer” (ImageJ 1.50b, National Institutes of Health) platelets, and JAQ1^Dy488 “clusters” were quantified.

Results from at least 3 experiments per group are presented as mean ± standard deviation. Differences between 2 groups were assessed by Welch t-test. *P < .05; **P < .01; ***P < .001.

As reported previously,^10,11  the anti-GPVI antibody, JAQ1, induced a rapid loss of platelet surface GPVI and the appearance of soluble GPVI (sGPVI) in the plasma, which was accompanied by a transient thrombocytopenia (see supplemental Figure 1, available on the Blood Web site). In sharp contrast, JAQ1-F(ab′)₂ fragments neither caused thrombocytopenia (supplemental Figure 1A) nor elevated plasma sGPVI levels (supplemental Figure 1C), demonstrating that both processes are Fc-dependent. A potential involvement of FcγRs was assessed by treating mice with the antibody 2.4G2, which blocks FcγRIII and IIB,¹⁹  24 hours before JAQ1 injection. The 2.4G2 abolished both anti–GPVI-induced thrombocytopenia and generation of sGPVI (Figure 1). To determine which of the 2 FcγRs was required for the anti-GPVI effects, we analyzed Fcgr3^−/− and Fcgr2b^−/− mice. Surprisingly, although a lack of FcγRIII had no effect on JAQ1-induced thrombocytopenia or generation of sGPVI (Figure 1D-F), both processes were abolished in FcgR2b^−/− mice, resulting in delayed GPVI downregulation (Figure 1E), presumably reflecting internalization/degradation.¹⁰ 

Anti-GPVI opsonized platelets accumulated in the liver of wild-type (WT), but not of FcgR2b^−/−, mice (Figure 2A-D). Remarkably, only one-third of the “trapped” WT platelets was in close proximity to Kupffer cells (Figure 2D, green arrows), whereas virtually all platelets were found close to the endothelium. Kupffer cell depletion with clodronate liposomes²⁰  had no effect on anti–GPVI-induced thrombocytopenia, platelet accumulation in the liver, or sGPVI plasma levels (Figure 2E-F; supplemental Figure 2). Furthermore, we detected robust signals for clusters of JAQ1^Dy488 in liver sections of WT, but not FcgR2b^−/− mice (Figure 2G-H and data not shown), which likely reflected large sGPVI-JAQ1 complexes. Notably, some of the JAQ1^Dy488 signal colocalized with platelets (Figure 2H, blue arrows), whereas clusters of nonplatelet-associated antibodies were also found within endothelial or Kupffer cells (Figure 2H), likely representing sGPVI-JAQ1 complexes released from the platelet surface.

Analysis of bone marrow chimeric mice lacking FcγRIIB either in the hematopoietic or in the nonhematopoietic compartment (ie, the endothelium) demonstrated that the absence of endothelial FcγRIIB ameliorated the anti–GPVI-induced thrombocytopenia and resulted in a delayed loss of surface GPVI and reduced generation of sGPVI (supplemental Figure 3B-C) independently of the expression of the receptor in hematopoietic cells (supplemental Figure 3A).

Our results reveal a novel role of the inhibitory FcγRIIB in processing anti–GPVI-opsonized platelets. Based on the results obtained in Kupffer cell depleted and in bone marrow chimeric mice, we conclude that LSECs are the critical cell type to execute this function. The (transient) thrombocytopenia promoting effect of FcγRIIB was unexpected because this receptor is best known for its inhibitory action on either activating FcγRs or B-cell receptors.²¹  Given the differences in FcγR expression between mice and humans,¹²  further studies will be required to assess the exact nature and potential IgG subclass specificity of FcγRs mediating GPVI-immunodepletion in humans.

It is unclear at present how the interaction between FcγRIIB and antibody-opsonized platelets would result in the quantitative loss of platelet GPVI. However, the high abundance of FcγRIIB on LSECs and its capacity to cluster anti-GPVI antibodies likely provides a sufficient cross-linking array necessary for platelet binding and activation that in turn results in the generation of sGPVI. Combined loss of ADAM17 and ADAM10 in platelets prevents GPVI shedding in vitro, but not in vivo.¹¹  Shedding in trans may occur, because it has been described for ADAM10-mediated processing of ephrin-5a.²²  However, the production of sGPVI in mice constitutively lacking ADAM17¹¹  or ADAM10 in endothelial cells (supplemental Figure 4) was unaltered. Thus, the identity of the protease(s) responsible for the generation of sGPVI in vivo remains to be determined.

Taken together, our data reveal for the first time an unexpected function of LSEC-FcγRIIB acting in trans for the targeted GPVI ectodomain shedding in vivo, at least in mice. These results may have an impact on the development of antibody-based anti-platelet therapies.

The authors thank Juliana Goldmann and Stefanie Hartmann for excellent technical assistance, Deya Cherpokova for critical comments, and the Bio-Imaging Center for providing technical infrastructure.

This work was supported by the Deutsche Forschungsgemeinschaft (SFB 688) and the Rudolf Virchow Center.

Contribution: D.S., M.P., V.L., and J.K.W. performed experiments and analyzed data; D.S. and J.E.G. interpreted results and contributed to the writing of the manuscript; J.E.G. provided vital reagents; and B.N. designed research, funded the project and wrote the manuscript.

Conflict-of-interest disclosure: The authors declare no competing interests.

Correspondence: Bernhard Nieswandt, Institute of Experimental Biomedicine, University Hospital Würzburg and Rudolf Virchow Center for Experimental Biomedicine, University of Würzburg, Josef-Schneider-Str 2, 97080 Würzburg, Germany; e-mail: bernhard.nieswandt@virchow.uni-wuerzburg.de.