An acquired inhibitor to the GPVI platelet collagen receptor in a patient with lupus nephritis

Authors


Alan T Nurden, Centre de Référence des Pathologies Plaquettaires, Plateforme Technologique et d’Innovation Biomédicale, Hôpital Xavier Arnozan, 33604 Pessac, France.
Tel.: +33 5 57 10 28 51; fax: +33 5 57 10 28 64.
E-mail: alan.nurden@cnrshl.u-bordeaux2.fr

Abstract

Summary. Background: GPVI is a major platelet collagen signaling receptor. In rare cases of immune thrombocytopenic purpura (ITP), autoantibodies to GPVI result in receptor shedding. Objectives: To investigate a possible pathogenic role of plasma anti-GPVI antibody located in a woman with lupus nephritis. Methods: Measured were (i) platelet aggregation to collagen and convulxin, (ii) platelet GPVI expression (flow cytometry and western blotting), (iii) plasma soluble GPVI (sGPVI, dual antibody ELISA), and (iv) plasma anti-GPVI antibody (ELISA using recombinant sGPVI). Results: In 2006 and early 2007, the patient had a normal platelet count but a virtual absence of platelet aggregation to collagen and convulxin. Her platelets responded normally to other agonists including cross-linking ITAM-dependent FcγRIIA by monoclonal antibody, IV.3. Flow cytometry and western blotting showed a platelet deficiency of GPVI. Plasma sGPVI levels were undetectable whereas ELISA confirmed the presence of anti-GPVI antibody. Sequencing revealed a normal GPVI cDNA structure. The patient’s plasma and the isolated IgG3 fraction activated and induced GPVI shedding from normal platelets. A deteriorating clinical condition led to increasingly strict immunosuppressive therapy. This was globally associated with a fall in plasma anti-GPVI titres, the restoration of platelet GPVI and the convulxin response, and the loss of her nephrotic syndrome. Conclusions: Our results show that this patient acquired a potent anti-GPVI IgG3 antibody with loss of GPVI and collagen-related platelet function. Further studies are required to determine whether anti-GPVI antibodies occur in other lupus patients with nephritis.

Introduction

Collagen is a major thrombogenic constituent of the vessel wall being both a substrate for platelet adhesion and a potent stimulator of platelet aggregation [1]. Platelets principally react with collagen through two receptors, integrin α2β1 and GPVI, a member of the immunoglobulin receptor family [2,3]. The platelet response involves a multi-step activation pathway with a major role played by GPVI through signaling mediated by the non-covalently associated immunoreceptor tyrosine-based activation motif (ITAM)-containing FcRγ-chain. Loss of GPVI receptor function has been associated with autoantibodies to GPVI in immune thrombocytopenic purpura (ITP) [4–7]. These antibodies induce receptor shedding, a phenomenon first observed in mice injected with rat monoclonal antibodies (MoAbs) to murine GPVI where a loss of platelet response to collagen unexpectedly continued for at least 2 weeks after their injection [8,9]. The depletion appears to be mediated by ADAM(a disintegrin and metalloprotease)10-induced cleavage with release of soluble GPVI ectodomain (sGPVI) [10,11]. Immune-induced endocytosis with surface clearance of GPVI may also contribute to a loss of the collagen response [12].

Systemic lupus erythematosus (SLE) is an autoimmune disease associated with the development of anti-DNA and, in many patients, anti-phospholipid antibodies [13]. While antibodies against platelet receptors have been described in SLE [14], we have found only one report of an antibody to GPVI in this disease [15]. Our study concerns the characterization of an IgG3 autoantibody against GPVI in a patient with lupus nephritis whose platelets transiently failed to respond to collagen and lacked GPVI. In short-term incubations, the purified antibody was able to activate normal platelets through a GPVI-dependent mechanism while prolonged incubation resulted in GPVI loss and Cvx unresponsiveness. Using ELISA and other assays, we showed that immunosuppressive therapy was accompanied by a fall in circulating anti-GPVI antibody levels, the restoration of both platelet GPVI and the platelet aggregation response, and an improved clinical condition.

Materials and methods

Case report

A 22 year-old woman of Moroccan nationality presented early 2005 with severe thrombocytopenia (4000 platelets μL−1), a bleeding diathesis and a suspicion of Evans syndrome. Marrow biopsy showed a rich megakaryocyte content; her Coombs test was positive and C4 complement levels low leading to a diagnosis of ITP. A weak presence of anti-nuclear antibodies was noted but tests for anti-DNA were negative. She responded to corticoids and, by May 2005, her platelet count had restored to 250 000 platelets μL−1. In February 2006, the onset of a lupus-like nephropathy led to the diagnosis of SLE; tests for DNA antibody were now positive but antibodies to phospholipids were not present and have since remained negative. She was treated with mycophenolate mofetil (MMF) and corticoids. In January 2007, the patient received platelet transfusions prior to a lumbar puncture during tests to explain diplopia and severe headaches. Tolosa-Hunt syndrome, painful ophthalmoplegia caused by inflammation of the cavernous sinus or superior orbital fissure was diagnosed. By the summer of 2007, she had developed a nephrotic syndrome with strong anti-DNA antibody titers. In May 2007, she received platelet transfusions prior to a kidney biopsy; however, this was aborted as a result of a failure of the transfused platelets to restore the collagen response. In May to December 2007, treatments included increased doses of corticoids while other immunosuppressive therapy included azothioprine for 2 months, and finally 6-monthly treatments with cyclophosphamide (1 g per bolus). In early 2008, MMF was reinitiated as maintenance therapy. By the summer of 2008, anti-DNA antibody titers had decreased significantly. Her kidney condition has improved with the disappearance of the nephrotic syndrome. Since November 2006, her platelet counts have ranged from 309 000 to 441 000 μL−1, her platelet volume is normal. During this time, she has not suffered from bleeding. Written informed consent was obtained according to the Declaration of Helsinki and the studies performed with ethical authorization given to the French Network ‘GIS-Maladies Rares’.

Platelet function testing

Platelet aggregation was tested in citrated platelet-rich plasma (PRP) at 250 000 platelets μL−1 using 0.5 and 2 μg μL−1 Horm equine tendon collagen (Nycomed Pharma, Unterschleiβheim, Germany), 400 or 800 pm convulxin (Cvx; provided by M-JP), 5 and 10 μm ADP (Sigma–Aldrich Chimie, Lyon, France), 500 μg mL−1 arachidonic acid (AA; Nu Chek Prep, Elysian, MN, USA), 4 μm epinephrine (Sigma–Aldrich), 50 μm thrombin receptor activating peptide (TRAP; Neosystem SA, Strasbourg, France), and 0.5 and 1.5 mg mL−1 ristocetin (Stago, Asnières-sur-Seine, France) in a PAP4 aggregometer (Bio/Data Corp., Wellcome Lab., Paris, France) according to standard procedures. ITAM-dependent signaling through phospholipase Cγ2 was tested using washed platelets incubated with 5 μg mL−1 of the MoAb, IV.3 (Medarex, Annandale, NJ, USA) and F(ab’)2 fragments of an antibody to mouse IgG (Eurobio, Les Ulis, France) as previously described [16]. The PFA-100 (Bayer, Puteaux, France) closure time was assessed using collagen/epinephrine cartridges according to the manufacturer’s protocol.

Flow cytometry

According to our standard procedures, surface receptors were analyzed using platelets in citrated PRP using MoAbs to αIIbβ3 (AP2, from Dr. Kunicki, La Jolla, CA, USA), GPIbα (Bx-1, produced in Bordeaux), α2β1 (Gi9; Beckman-Coulter, Marseille, France) and GPVI (3J24.2, from M J-P) [16,17]. FcRγ-chain was assessed using platelets fixed in 1% paraformaldehyde (PFA) and permeabilized using 0.1% Triton X-100 for 30 min at 4 °C followed by the addition of purified IgG (5 μg mL−1) of a rabbit antibody (Euromedex, Mundolsheim, France) [16]. Incubation of platelets for 10 min with 50 μm TRAP was followed by analysis of the surface expression of P-selectin using VH10 (produced in Bordeaux). Bound primary antibodies were detected using FITC-labeled F(ab’)2 fragments of sheep antibody specific for mouse or rabbit IgG (Eurobio). Controls were isotype-specific mouse IgG or non-immune rabbit IgG. Platelets were analyzed in a Cytomics FC500 flow cytometer (Beckman Coulter, Villepinte, France) and results expressed as mean fluorescence intensity (MFI). Histograms were generated from measurements of 10 000 cells and data analyzed using CXP software (Beckman Coulter).

Western blotting

Washed platelets from control donors or the patient were lysed in buffer containing 2% sodium dodecyl sulfate (SDS) according to our standard procedures [16]. On occasion, platelets were also solubilized with Triton X-100 as for ELISA (see below). Sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using 10%, 15% or 4–12% gradient gels (see Figures 2, 3 and 7) and transferred to nitrocellulose or PVDF membrane. Selected platelet membrane glycoproteins (GPs) were assessed using the MoAbs Bx-1 (GPIbα), SZ22 (αIIb, Beckman Coulter), Y2/51 (β3, Dakopatts, Glostrup, Denmark), VH10 (P-selectin) and a rabbit antibody to TLT-1 (from Dr. D. MacVicar, NIH, Bethesda, USA) [16,18]. GPVI was evaluated using MoAbs to the extracellular domain (TK.5, from Dr T. Kunicki; OM1 prepared in Rockville by N.T.; 3J24.2 from M.J-P. and human polyclonal anti-GPVI from Dr M. Okuma, Kyoto, Japan) [5,16,17,19]. Also used, was a rabbit antibody to the GPVI intracytoplasmic domain (from Dr. M. Berndt, Melbourne, Australia) [7] and a commercial rabbit antibody to the FcRγ subunit [16]. OM1 was conjugated directly to horseradish peroxidase (HRP) using a HRP-labeling kit as described by the manufacturer (Dojindo Molecular Technologies, Rockville, MD, USA); the other MoAbs were detected using peroxidase-linked anti-mouse or anti-rabbit IgG. Bound antibody was detected by chemiluminescence (ECL-Amersham Biotech, Little Chalfont, UK).

Preparation of recombinant sGPVI

The methodology for the preparation of monovalent sGPVI was adapted from that described by Miura et al. [20]. The first step was the establishment of a cell line expressing His6 and V5 tagged sGPVI. For this, the extracellular domain of human GPVI encoding 22nd–219th amino acid residues was cloned into the plasmid pEF1/SecTagV5/HisA (Invitrogen) using EcoRI/NotI. The resulting plasmid was transfected into CHO-K1 cells using Lipofectin (LF2000, Invitrogen) according to the manufacturer’s instructions. Finally, single stable clone-producing sGPVI was selected as G418 resistant cells. This was followed by the production and purification of sGPVI. sGPVI-expressing CHO cells were cultured in DMEM/F12(1:1) containing 2.5% FCS and 100 μg mL−1 G418 at 37 °C in a 5% CO2-containing atmosphere. The cells were cultured in roller bottles until they reached confluence. Supernatants were collected, centrifuged (3800 g, 30 min, 4 °C) to remove cell and debris, filtrated through a (0.2 micron) filter and concentrated by Pellicon XL (with a 10-kDa filter; Millipore). sGPVI was purified by affinity chromatography on anti-GPVI MoAb (OM1)-Sepharose matrix. Briefly, OM1[19]-affinity matrix was prepared by coupling OM1 IgG to NHS HP-columns [HiTrap(tm) NHS HP, Amersham Biotech] as described by the manufacturer. The concentrated supernatant was loaded onto a OM1-Sepharose column equilibrated in phosphate-buffered saline (PBS). The column was washed with PBS until the OD280 was < 0.01 and the bound sGPVI was eluted with 3M KSCN in PBS. The eluted fractions were pooled, dialyzed against PBS, concentrated and frozen at −20 °C until used. Recombinant s(GPVI)2Fc was prepared as described with the exception that purification was now performed on Protein-A Sepharose and Nickel columns [2,20].

Measurement of anti-GPVI antibodies in plasma

These tests were performed using an ELISA developed by Otsuka Maryland Medicinal Labs (Rockville, MD, USA). Briefly, recombinant monovalent sGPVI was coated onto a 96-well plate followed by blocking of non-specific binding sites. Serially diluted plasma samples were added to capture anti-GPVI antibodies that were detected with goat anti-human IgG [F(ab’)2] labeled with peroxidase (American Qualex, San Clemente, CA, USA).

Measurement of sGPVI in plasma by sandwich ELISA

Murine MoAbs OM1 and OM2 were produced by immunization of GPVI knockout mice with CHO cells stably expressing human GPVI [19]. OM1 (the capture antibody) was coated onto a 96-well plate. After blocking, serially diluted plasma was added and bound sGPVI detected using biotinylated OM2-Fab and peroxidase-labeled streptavidin (Zymed, San Francisco, CA, USA). Biotinylated OM2 was prepared using the biotin labeling Kit-NH2 as suggested by the manufacturer (Dojindo Molecular Technologies). Standard curves were constructed using soluble recombinant sGPVI. It should be noted that OM1 and OM2 recognize distinct sites on GPVI and that their binding is not competitive [19].

DNA analysis

Genomic DNA was prepared from peripheral blood leukocytes from the patient and control donors and all eight exons of the GP6 gene together with splice sites were amplified by polymerase chain reaction (PCR) and sequenced using standard protocols. The structure of the oligonucleotides is available on request.

Isolation and characterization of the patient’s antibody

Samples (500 μL) of plasma from the patient or control donors were applied to mini-columns of protein A Sepharose (HiTrap; GE Healthcare, Saclay, France). Bound IgG was eluted using 100 mm glycine HCl, pH 3, followed by dialysis against PBS (pH 7.2). The flow-through was retained and treated similarly. Volumes (10–50 μL) of the patient’s plasma obtained at different dates were added to 250 μL of normal citrated PRP before stimulation by collagen or TRAP. Alternatively, the purification fractions were mixed with an equal volume of washed platelets (2.5 × 108 mL−1) [21]. Platelet aggregation and P-selectin expression were assessed according to the procedures described in above sections. As a positive control, platelets were incubated with the GPVI activating MoAb 9012 [22]. The binding specificity of fractions isolated from the patient’s plasma was tested in ELISA using recombinant sGPVI coated at 2 μg mL−1 in microtiter plates. IgG isotyping was performed by the same procedure using MoAbs specific for IgG1 (NL16), IgG2 (GOM2), IgG3 (ZJ4) and IgG4 (RJ4) (from Dr Pierre Aucouturier, Hôpital Tenon, Paris, France). For in vitro analysis of GPVI shedding, washed control platelets (5 × 108 mL−1) were mixed with an equal volume of the non-retained fractions or purified IgG at 37 °C for 20 min prior to the addition of 5 mm EDTA, centrifugation and preparation of the platelets for SDS-PAGE and western blotting using the anti-GPVI MoAb 3J34.2 as described in an earlier section.

Results

Initial demonstration of a platelet GPVI deficiency

Results of studies performed on the patient in November 2006 revealed a prolonged PFA-100 closure time (collagen/epinephrine cartridge, 242 s; control range 50–150 s) and little or no platelet aggregation at even a high dose of collagen (Fig. 1A). In contrast, aggregation was normal to agonists such as ADP and ristocetin (Fig. 1A) or AA, epinephrine and TRAP (not illustrated). The specific nature of the collagen defect led us to test Cvx, a snake venom protein that binds to GPVI [2]. The patient’s platelets failed to respond to 400 pm Cvx (Fig. 1A), increasing the dose did not restore the aggregation (not shown). However, incubation of the patient’s platelets with the MoAb, IV.3 cross-linked with anti-mouse IgG was followed by a full-scale aggregation indicating normal FcγRIIA and ITAM-dependent signaling (Fig. 1B) (see Fig. 6 of ref (16) for an identical response of control platelets). Flow cytometry showed that the patient’s platelets normally expressed GPIbα, αIIbβ3 and α2β1 (Fig. 1C); while platelet stimulation with TRAP resulted in the normal expression of P-selectin (not shown). Nevertheless, results with the MoAb 3J24.2 suggested a severe deficiency of GPVI on the patient’s platelets (Fig. 1C). In order to eliminate a down-regulation with the accumulation of internal pools [12,23], we performed flow cytometry using PFA-fixed and permeabilized platelets. The results confirmed that total GPVI was decreased in the patient’s platelets; the non-covalently associated FcRγ chain was also decreased but less so (data not shown).

Figure 1.

 Collagen-induced platelet aggregation and GPVI expression in platelets isolated from the patient in November 2006. (A) Platelet aggregation performed in citrated platelet-rich plasma (PRP) showed an absent response to 20 μg mL−1 collagen and 400 pm Cvx whereas the response to 1.5 mg mL−1 ristocetin and 10 μm ADP was retained. (B) The platelets also rapidly aggregated when incubated with 5 μg mL−1 MoAb IV.3 and F(ab’)2 fragments of an antibody to mouse IgG. (C) Flow cytometry confirmed a normal receptor expression on the patient’s platelets with the exception of GPVI which decreased significantly. % LT, light transmission; min, minutes; MFI, mean fluorescence index.

Seen as a ∼ 58-kDa band in western blotting, GPVI was severely reduced in the platelets of the patient with the FcRγ chain which also diminished but less so (Fig. 2). Results are shown for the MoAb 3.J24.2, similar profiles were seen with TK.5 and OM.1 all recognizing the extracellular domain while a decreased GPVI was also seen with the human polyclonal anti-GPVI. With a rabbit antibody to the GPVI intracytoplasmic domain, the 58-kDa GPVI band was again decreased in the patient’s sample. A band of ∼ 45 kDa consistently seen with this antibody for both patient and control samples is of unknown origin. Other platelet membrane glycoproteins, including GPIbα, P-selectin and TLT-1, all previously reported substrates for platelet metalloproteases [18], were normally present and also served as loading controls for the anti-GPVI filter (data not shown). Sequencing of genomic DNA revealed no potential pathological mutations in the coding or splicing regions of the GP6 gene that could account for the absence of the protein from the platelets; examination of GP6 SNPs showed that the patient belonged to the SKTQH phenotype [24,25].

Figure 2.

 Western blotting of GPVI in platelets isolated from the patient in November 2006. Serial dilutions of proteins in sodium dodecylsulfate (SDS)-platelet lysates (non-reduced) from the patient or a control donor were subjected to separation on 10% or 15% gels as shown. After transfer to nitrocellulose membrane, GPVI or FcRγ were revealed using a mouse MoAb to GPVI (3.J24), or rabbit antibodies to FcRγ or the GPVI intracytoplasmic domain. Chemiluminescence detection of bound IgG revealed a major deficiency of GPVI in the patient’s platelets and a partial deficiency of FcRγ. A band of ∼ 45 kDa detected by the rabbit antibody to the GPVI intracellular domain is of unknown origin.

Evolution of GPVI expression in the patient’s platelets with treatment

During the development of the patient’s kidney disease, platelets have been regularly tested and also the different immunosuppressive therapies. As shown in Fig. 3, GPVI-dependent aggregation to Cvx was still at basal levels in May 2007 but had improved by September 2007 (panel A) when the PFA-100 closure time had normalized (data not shown). Flow cytometry (data not shown) and western blotting (panel B) confirmed that GPVI levels in the patient’s platelets increased in parallel with their capacity to respond to Cvx. Results obtained using the MoAb, OM-1, while confirming a reduced GPVI expression in a sample obtained in May 2007, shows that GPVI was readily detectable in samples taken in October 2007 and in January 2008. GPVI was present in lysates prepared in both the ionic detergent SDS and the nonionic detergent Triton X-100. OM1 also recognized the recombinant sGPVI used in ELISA (data not shown).

Figure 3.

 Evolution of the platelet convulxin response and GPVI content as tested during treatment. (A) Platelet aggregation in response to 800 pm convulxin was tested at the dates shown: azothioprine was given in June 2007 followed by five monthly injections of cyclophosphamide beginning in July 2007. (B) Western blotting with the MoAb OM1 was also performed. Patient samples (20 μg protein) were (i) in sodium dodecylsulfate (SDS) lysate (05/07), (ii) Triton X-100 lysate (10/07), (iii) SDS lysate (01/08) and (iv) Triton X-100 lysate (01/08). Also shown is a Triton X-100 lysate from control platelets (20 μg protein).

Evaluation of circulating anti-GPVI antibodies and GPVIs

During treatment, the reappearance of GPVI in the patient’s platelets strongly suggested that the initial deficiency was acquired. Using a newly developed ELISA, we retrospectively analyzed plasma from the patient for anti-GPVI antibodies. This test used recombinant sGPVI in the microtiter wells. As shown in Fig. 4, anti-GPVI antibodies were strongly present at a time when the patient responded poorly to convulxin (February 2007). However, in plasma taken in May 2007 and thereafter, they had decreased greatly. We also tested for sGPVI in the patient’s plasma. Here, the capture ELISA used two MoAbs that react with different and non-competing epitopes on GPVI. The MoAb, OM1, retained sGPVI in the microtiter wells, whereas biotinylated OM2 allowed a measure of the captured soluble receptor. Corrections were not made for changes in platelet count. Studies on 10 normal donors found an average of 9.2 ± 3.3 ng sGPVI mL−1 plasma (range 6–12 ng mL−1). Figure 5 shows that sGPVI was not detected in a sample taken from the patient in February 2007 and was present at greatly reduced levels in May 2007. In contrast, sGPVI was normally present in plasma from the patient from September 2007, its presence paralleled platelet GPVI expression (refer back to Fig. 3). Interestingly, plasma sGPVI was transiently increased after platelet transfusions given in May 2007 whereas plasma antibody levels were not affected.

Figure 4.

 A direct ELISA was used to evaluate antibody to GPVI in the patient’s plasma. Samples were taken from the patient at the dates shown or from three control donors (Cont) and analyzed in duplicate using two plasma dilutions and recombinant sGPVI adsorbed within the microtiter wells. Bound human IgG was detected with a peroxidase-labeled goat anti-human IgG [F(ab’)2]. Also shown are results for plasma samples taken in May 2007 when the patient twice received platelet transfusions (see Case History in Methods). BT, before transfusion; AT, after transfusion; OD, optical density.

Figure 5.

 A two-antibody capture ELISA was used to evaluate sGPVI levels in the patient’s plasma. Samples were taken from the patient or from three control donors (Cont) at the dates shown and as detailed in the legend to Fig. 4. The results are expressed as ng mL−1 sGPVI. Note that sGPVI levels were absent in February 2007 present at low amounts in May 2007 and increased after the platelet transfusions. Levels were normalized by September 2007.

Identification of an activating plasma IgG3 antibody to GPVI

Plasma taken when the patient’s GPVI levels were low was able to activate control platelets and induce a rapid dose-dependent aggregation (Fig. 6). This aggregation was blocked in a concentration-dependent manner by recombinant s(GPVI)2Fc [20] but not by MoAb, IV.3. Significantly, when this same plasma was incubated with unstirred control platelets for longer periods, spontaneous aggregation was abolished and there was a major reduction in platelet aggregation by 400 pm convulxin (maximal intensity 25% vs. 77% after incubation with control plasma). The response to 50 μm TRAP changed little (66% vs. 75%). To see if this activity could be produced by purified antibody, plasma from the patient was passed through a Protein A-Sepharose column. To our surprise, the activating activity was present in the flow-through fraction. Only the latter was able to induce aggregation and P-selectin exposure of control platelets (data not shown). This was despite IgG recognizing recombinant sGPVI being detected in both the non-retained and eluted fractions after the Protein A-Sepharose chromatography (Fig. 7A). Anti-GPVI antibody was not present in either fraction of control plasma. Immunotyping revealed that the antibody in the non-retained fraction of the patient’s plasma belonged to the IgG3 class. Finally, the ability of the non-retained and eluted fractions to induce GPVI cleavage in vitro was compared with that of the activating anti-GPVI MoAb, 9012.2. As shown in Fig. 7B, the flow-through fraction (IgG3) brought about a complete loss of GPVI from control platelets as did 9012.2. Western blotting confirmed that sGPVI was now present in the supernatant confirming that the IgG3 antibody was inducing shedding (not shown).

Figure 6.

 The patient’s antibody is able to induce platelet activation. Plasma from the patient taken in November 2006 (PP1, 50 μL) induced a rapid aggregation when added to stirred platelet-rich plasma (PRP) from a control donor. This activity was not present in plasma taken in January 2008 (PP2). The aggregation was inhibited by the parallel addition of recombinant s(GPVI)2Fc (VIFc, 80 μg mL−1) but not by the MoAb IV.3 (10 μg mL−1).

Figure 7.

 Studies using purified IgG fractions. Plasma from the patient or a control donor was applied to a Protein A-Sepharose column and both the non-retained flow-through and IgG eluate fractions were tested for their ability to bind to recombinant sGPVI or BSA in ELISA (A) or induce GPVI cleavage (B). In (A), bound IgG was detected using an HRP-coupled anti-human IgG (Ab1). Note that antibody to GPVI was present in both fractions. In (B), aliquots from both the non-retained and eluted fractions or purified activating 9012.2 IgG were incubated with washed control platelets for 20 min at 37 °C prior to the analysis of platelet GPVI content by western blotting using the MoAb 3J24.2. Shown are platelets analyzed without incubation (no inc), platelets incubated with phosphate-buffered saline (PBS) and platelets incubated with control (Cont) or patient’s plasma (Pat) or with 9O12.2 IgG.

Discussion

We describe a woman who in 2005 had an ITP-like syndrome that was treated with corticoids. Subsequently, the onset of kidney disease and the detection of anti-DNA antibodies led to the diagnosis of lupus nephritis. Although her platelet count had normalized, platelet function testing showed her platelets to be unresponsive to collagen and Cvx, thus suggesting an abnormality in the GPVI signaling pathway. While isolated cases with abnormal signaling through GPVI have been described [26–28] and preliminary reports suggest the existence of inherited defects in the GP6 gene leading to loss of GPVI expression and/or function [29,30], the bulk of reports from mouse models or human disease suggest that a more common cause of non-responsiveness to collagen is GPVI loss linked to antibody-modulated sheddase activity [6–11,31,32]. As sequencing of the coding regions and splice sites of the GP6 gene of our patient failed to reveal mutations indicative of a genetic defect, and that the patient was homozygote for the the common SKTQH GPVI haplotype said to be associated with a higher GPVI function [24,25], our efforts turned towards searching for an acquired defect and antibody promoted GPVI cleavage.

Western blotting using a battery of antibodies to different epitopes on GPVI identified a specific deficiency of GPVI in the patient’s platelets. Quantifying antibody binding to recombinant sGPVI by ELISA confirmed the presence of anti-GPVI antibodies in the plasma taken at this time. This antibody interacted in vitro with control platelets in two phases, an immediate activating action with platelet aggregation, and a slower time-dependent loss of the platelet response to collagen. The activating activity was blocked by recombinant s(GPVI)2Fc but not by the MoAb IV.3 that prevents Fc-mediated IgG binding to FcRγIIA [33]. Several MoAbs to GPVI have been shown to activate platelets and both FcRγIIA-dependent and -independent mechanisms have been described [19,22,23,32,34,35]. One possible pathway for FcRγIIA-independent activation is the induction of disulfide-dependent dimerization of GPVI in the plasma membrane with loss of calmodulin binding to the intracellular domain, an activating pathway used by Cvx [36]. It is significant that the patient had a normal platelet count and no severe bleeding at the time that her antibody titers were high, yet her platelets had a virtual complete loss of GPVI. Similar findings have been reported for other GPVI-deficient patients [5,6,26] and for GPVI-deficient mice [37]. In contrast, when rat MoAbs of the denominated JAQ family were infused into mice a rapid and quantitative down-regulation of GPVI was accompanied by a transient thrombocytopenia [8,9]. For our patient, we do not know if anti-GPVI antibodies were present during the severe thrombocytopenia observed in 2005. Also unknown is the true relevance of the IgG3 activating antibody to GPVI in the development of the nephritis that characterized the lupus of the patient. IgG3 subtype antibodies are characterized by the presence of a prolonged hinge region and the distance between the Fc- and active-site domains [38]. Nevertheless, it is of interest that IgG3 has been reported to be an important factor in the pathogenesis of this particular manifestation of this disease [39].

One aspect of our work needs underlining. The development of ELISA tests for plasma antibodies to GPVI and sGPVI allowed us to compare levels of both items. But these analyses were performed retrospectively on samples stored during the evolution of her disease and results were therefore not a factor in determining treatment. The use of increasingly severe immunosuppressive therapy was based on the clinical evolution of the patient’s condition with emphasis on (i) her kidney disease and (ii) increasing titres of anti-DNA antibodies. Interestingly, we failed to detect sGPVI in her plasma when platelet GPVI levels were low and plasma anti-GPVI antibody levels high. Boylan et al. made a similar observation for an ITP patient [6]. While endocytosis may contribute to the phenomenon, WB blotting clearly pointed to GPVI cleavage as the major cause. Neverthless, GPVI shedding in another case of ITP led to increased levels of sGPVI [7]. The reasons for this variation are unknown although they could relate to antibody titre and the ability of the antibody to react with MKs and induce GPVI loss in the marrow. It is also possible that high titer plasma antibody leads to the formation of circulating immune complexes that would be cleared. As time progressed, her levels of sGPVI normalized and this correlated well with the increase in platelet GPVI levels. The critical period was May 2007 when the platelet response to collagen remained low and platelet GPVI content minimal. Nevertheless, plasma sGPVI levels were already on the upward trend and plasma antibody levels had decreased. This suggested that the patient’s corticotherapy was having an effect prior to the start of azothioprine. Interestingly, these results can be likened to those reported by Nieswandt’s group [8,9] who observed that infusing rat anti-GPVI MoAbs into mice resulted in inhibition of the platelet collagen response for well beyond the expected lifespan of the antibody in the circulation. This finding has yet to be explained but could result from a secondary effect of the antibodies in the marrow, with GPVI shedding and even a down-regulation of GPVI synthesis. Such factors could contribute to the absence of sGPVI in the patient’s plasma in early 2007.

Our study raises an important question. Is the activating IgG3 antibody to GPVI linked to the progression of the lupus pathology developed by the patient in early 2007? The presence of anti-DNA antibodies and the development of nephritis are typical of a subgroup of lupus patients. Our patient consistently tested negatively for anti-phospholipid and anti-beta2 glycoprotein I antibodies found in many lupus patients [40]. The presence of an activating IgG3 antibody able to induce activation-dependent GPVI shedding (a mechanism proposed in 24) suggests that the structurally different IgG3 can more efficiently cross-link GPVI in the plane of the membrane than other isotypes of IgG. We are aware of only one previous report of antibodies to GPVI in a Japanese patient with lupus [17]. In view of our findings it would be interesting to screen a large series of lupus patients and particularly those with a modified collagen-induced platelet aggregation to determine the potential pro-nephritic(via immune complex formation) and anti-thrombotic (via GPVI loss and a down-regulation of the collagen response) consequences.

Addendum

P. Nurden and A.T. Nurden planned the study; N. Tandon and H. Takizawa performed ELISA assays for plasma sGPVI and anti-GPVI antibody; D. Morel and L. Couzi followed the patient clinically; X. Pillois and M. Fiore performed DNA analysis; S. Loyau and M. Jandrot-Perrus did western blotting for GPVI and the studies on the isolated antibody. A.T. Nurden, P. Nurden, N. Tandon and M. Jandrot-Perrus all contributed to writing the of the manuscript.

Acknowledgements

This work was performed in the context of the French National Reference Center for Platelet Disorders sponsored by the French Health Ministry. A.T. Nurden and P. Nurden also acknowledge financial support from the French GIS-Maladies Rares and INSERM (Grant N° ANR-08-GENO-028-03). MJ-P acknowledges funding from the Fondation de France (Grant N° 2007001960). We thank R. Combrié, A. Concepcion and X. Gong for excellent technical assistance and P. Aucouturier for providing isotyping antibodies.

Disclosure of Conflict of Interests

A.T. Nurden is a member of the European Platelet Academy sponsored by Daiichi-Sankyo and Lilly. H. Takizawa and N. Tandon are research scientists at Otsuka Maryland Medicine Laboratories, Inc. The authors declare no competing financial interests.

Ancillary