Exacerbation of autoantibody-mediated thrombocytopenic purpura by infection with mouse viruses

Acute immune thrombocytopenic purpura (ITP) is a common disease of suspected autoimmune origin that often develops in the course of a viral infection, especially in children.¹  In affected patients, platelets opsonized with autoantibodies are destroyed through phagocytosis by cells of the reticuloendothelial system, leading to clinical symptoms such as petechiae or, in more severe cases, bleedings. Widely different viruses, including Epstein-Barr virus, influenza, varicella zoster virus, cytomegalovirus, mumps, parvovirus, and rubella virus have been suspected as potential triggers of the disease.¹  Molecular mimicry between epitopes shared by infectious agents and platelets, followed by spreading of the autoimmune response to new determinants, has been proposed as an explanation for the etiologic role of viruses in ITP.^2,3  Interestingly, proinflammatory cytokines, such as γ-interferon (IFN-γ) have been found at higher levels in patients with ITP when compared with healthy individuals.^2,3  Although the role of viruses in the pathogenesis of ITP is not disputable, it is, however, unclear how viruses with very divergent antigenic determinants can induce a similar autoimmune response focused to platelet epitopes through molecular mimicry.

So far, very few animal models have been developed that allow for an experimental approach of ITP. Mechanisms of platelet destruction may be analyzed in mice by using polyclonal antiplatelet xenoimmune antibodies or monoclonal antibodies elicited in rats or in mice.⁴  (NZW × BXSB)F1 mice spontaneously produce antiplatelet autoantibodies, leading to thrombocytopenia in the context of a complex autoimmune syndrome dissimilar to organ-specific ITP.⁵  Production of antiplatelet autoantibodies through molecular mimicry may be studied by immunizing mice with rat platelets, but the resulting thrombocytopenia is at best moderate and very transient in normal mice.⁶  Here, we report on the effect of infection with the common mouse viruses lactate dehydrogenaseelevating virus (LDV) and mouse hepatitis virus (MHV), on platelet counts in mice with preexisting antiplatelet autoantibodies. Infection triggers a clinical disease similar to childhood acute ITP through mechanisms that do not rely on autoantibody production by molecular mimicry but rather on activation of phagocytic cells after secretion of IFN-γ.

Specific pathogen-free CBA/Ht, BALB/c nu/nu female and 129/Sv female mice were bred at the Ludwig Institute for Cancer Research by G. Warnier and used when 8 to 12 weeks old. IFN-γ receptor-deficient mice (G129), received courtesy of F. Brombacher (Max Planck Institute for Immunobiology, Freiburg, Germany), were reared in the same manner as the 129/Sv mice, from which they were initially derived by S. Huang and M. Aguet.⁷  Some mice were purchased from B&K Universal (Hull, United Kingdom). Spleens were surgically removed after anesthesia with intraperitoneal administration of ketamine hydrochloride (Imalgene 500; 1.25 mg/mouse; Merial Toulouse, France) and medetomidine hydrochloride (Domitor; 25 μg/mouse; Orion Pharma, Espoo, Finland), followed by atipamezole hydrochloride (Antisedan; 125 μg/mouse; Orion Pharma).

Mice were infected by intraperitoneal injection of approximatively 2 × 10⁷ 50% infectious doses (ID₅₀) of LDV (Riley strain; ATCC, Rockville, MD) in 200 μL saline or of approximately 1 × 10⁴ 50% tissue culture infectious doses (TCID₅₀) of MHV-A59 grown in NCTC 1469 cells.

Blood was collected from the retroorbital plexus of ether-anesthetized mice with the use of Unopette pipettes and kits (Unopette microcollection system; Becton Dickinson, Franklin Lakes, NJ). Platelets were counted by microscopy with an improved Neubauer hemacytometer (Marienfeld, Germany).

Polyclonal antiplatelet antibodies were obtained after subcutaneous immunizations of a rabbit every 2 weeks with approximately 10⁹ BALB/C mouse platelets, once with complete Freund adjuvant, then with incomplete Freund adjuvant. Immunoglobulin G (IgG) was isolated by treatment with acridine lactate (Rivanol; Chinosol, Seelze, Germany) and precipitation with half-saturated ammonium sulfate. F(ab′)₂ fragments were prepared by peptic digestion, followed by purification by chromatography on Ultrogel AcA 4-4 and precipitation with ammonium sulfate.⁸  Rat anti-mouse CD41 (integrin α_IIb chain) monoclonal antibody⁹  was obtained from BD PharMingen (San Diego, CA). F96A9K1 is a monoclonal antiplatelet IgM autoantibody and N1 a monoclonal antiplatelet IgG2a autoantibody derived from (NZW × BXSB)F1 mice. Control monoclonal antibodies were rat IgG1 (LO-DNP1 from LO/IMEX; Université catholique de Louvain, Brussels, Belgium), mouse IgG2a (A6202F4), and mouse IgM (M444) of irrelevant specificity. Human IgG was Gammagard (Baxter, Lessines, Belgium). In vivo administration of antibodies was by the intraperitoneal route.

Clodronate-containing liposomes were prepared following the method described by Van Rooijen.¹⁰  Briefly, 86 mg phosphatidylcholine (Sigma, Bornem, Belgium) at 100 mg/mL in chloroform was added to 8 mg cholesterol (Sigma) in 10 mL chloroform. After evaporation, the lipidic film was dispersed by rotation for 10 to 15 minutes in 10 mL phosphate-buffered saline (PBS) alone or containing 2.5 g clodronate (dichloromethylene-bisphosphonate; kindly given by Boehringer Mannheim GmbH, now Roche Diagnostics GmbH, Germany). Liposomes were sonicated for 3 minutes at 50 W, then washed with PBS, and resuspended in 4 mL PBS. Mice were treated by intravenous injection with 200 μL liposomes.

Statistical analysis was performed by using nonparametric unpaired Mann-Whitney test.

To determine the effect of a viral infection on experimental ITP, we administered LDV and rabbit anti-mouse platelet polyclonal antibodies to normal mice. As shown in Figure 1A, when administered alone, both the virus and the antibody, at the selected dose, induced a moderate thrombocytopenia in CBA/Ht animals, whereas in mice with other genetic backgrounds, such as BALB/C, there were no consequences (not shown). In contrast, a sharp drop in platelet counts was observed in mice that received both the virus and the antiplatelet antibody (P = .0286 for mice treated with antiplatelet antibody and LDV versus antibody or virus alone). A similar exacerbation of antiplatelet antibody-mediated thrombocytopenia was observed in mice infected with MHV 1 day before antiplatelet antibody administration (P = .0079 for mice treated with antiplatelet antibody and MHV versus antibody or virus alone; Figure 1B). Analysis of this effect of viral infection was performed after LDV inoculation in all subsequent experiments. LDV infection was also found to enhance antiplatelet antibody-mediated thrombocytopenia in BALB/c nu/nu mice (P = .0286 for mice treated with antiplatelet antibody and LDV versus antibody or virus alone; Figure 2B). This effect of the virus led to a clinical picture reminiscent of purpura developing in patients (Figure 2A). Numerous petechiae could indeed be detected on mice that had received both the virus and the antiplatelet antibody but not in animals that were only infected or that were treated with the antibody alone.

LDV similarly worsened the thrombocytopenia induced by a rat monoclonal antibody recognizing mouse CD41⁹  (platelet glycoprotein IIb; BD PharMingen, San Diego, CA) and by an IgG2a (N1) and an IgM (F96A9K1) monoclonal autoantibodies derived from (NZW × BXSB)F1 mice that spontaneously develop autoantibody-mediated thrombocytopenia (as shown for typical experiments in Figure 3A). Moreover, kinetics experiments indicated that the effect of LDV was transient (Figure 3B). A sharp enhancement of autoantibody-mediated thrombocytopenia was indeed observed in mice that had been infected the same day, and 1 or 2 days before antiplatelet antibody inoculation (P = .0079), but not in animals that were infected for 4 days or more. Infection 1 or 2 days after antiplatelet antibody administration also resulted in an increase of thrombocytopenia (data not shown).

To further analyze the mechanisms by which a viral infection may exacerbate antibody-mediated thrombocytopenia, we first determined which part of the immunoglobulin was required in this phenomenon. Administration of F(ab′)₂ fragments of antiplatelet polyclonal antibody was not followed by the development of disease, even in infected animals, in contrast to intact immunoglobulin (significant difference between infected mice that received antiplatelet IgG or F(ab′)₂: P = .0286; Figure 4). This suggests that phagocytosis is involved in LDV-enhanced thrombocytopenia. This hypothesis was tested by administration of clodronate-containing liposomes to LDV-infected mice, a treatment known to suppress macrophages with phagocytic activity¹⁰  and, therefore, to cure experimental autoantibody-mediated blood autoimmune diseases that are mediated by phagocytosis.^6,11,12  Administration of these clodronate-containing liposomes resulted in a complete prevention of the increased pathogenicity of both polyclonal and monoclonal antiplatelet antibodies observed in LDV-infected mice (significant difference for all antibodies between mice treated with PBS- and clodronate-containing liposomes: P = .0286; Figure 5A). The hypothesis that LDV enhances thrombocytopenia through Fc receptor-mediated phagocytosis was further supported by administration of total human IgG to mice. This treatment has previously been shown to inhibit in vivo Fc receptor-mediated phagocytosis of autoantibody-coated red blood cells and to subsequently decrease the resulting hemolytic anemia.¹³  Similarly, LDV-enhanced antibody-mediated thrombocytopenia was largely suppressed by administration of total human IgG (significant difference between infected mice that received antiplatelet antibody and treated with total human IgG, or untreated: P = .0286; Figure 5B). However, the major site of this increased phagocytosis of antibody-coated platelets was not the spleen, because administration of antiplatelet antibodies to LDV-infected splenectomized mice and mock-operated animals induced a thrombocytopenia that was not quite significantly different (P = .0571; Figure 6).

Because IFN-γ is a potent activator of macrophage function, a possible role for this cytokine in LDV-induced enhancement of antibody-mediated thrombocytopenia was investigated by infecting mice deficient for the receptor of this cytokine (G129). Although the thrombocytopenia induced by the virus alone was not significantly different in these animals and in their normal counterparts, the aggravating effect of LDV on the pathogenicity of antiplatelet antibodies was largely decreased in mice unable to respond to IFN-γ. This suppression of LDV-enhanced thrombocytopenia in mice deficient for IFN-γ receptor was observed after administration of both polyclonal anti-mouse platelet rabbit IgG (Figure 7A) and monoclonal rat anti-mouse CD41 antibody (Figure 7B).

Our results indicate that LDV, in addition to inducing a transient thrombocytopenia by itself, can dramatically enhance thrombocytopenia that is triggered concomitantly to the infection by antiplatelet antibodies. This effect of the virus on antiplatelet antibody-induced thrombocytopenia does not require T lymphocytes, because it is observed in nude mice. Therefore, it is not caused by new antibody production in response to infection, because such antibody responses are T-cell dependent and do not start before 4 days after infection,¹⁴  and antigenic mimicry cannot explain this virally induced platelet drop. Because LDV may enhance phagocytosis,¹⁵  it could be postulated that the virus induces thrombocytopenia by this mechanism. Indeed, treatment with clodronate-containing liposomes indicated that the enhancing effect of LDV on autoantibody-dependent thrombocytopenia involves phagocytosis exclusively. Moreover, inhibition of this thrombocytopenia with total IgG indicated that LDV-enhanced phagocytosis of opsonized platelets was at least partly mediated by Fc receptors expressed on activated macrophages. Which type of receptors is involved and whether infection leads to an increase of the expression of these receptors remains to be determined. Interestingly, the moderate and transient thrombocytopenia induced by LDV infection in the absence of antiplatelet antibody administration was only partially prevented by treatment with clodronate-containing liposomes (data not shown). This suggests that different mechanisms may be involved in the thrombocytopenia induced by the virus alone or in the presence of antiplatelet antibodies.

Because macrophage functions, and especially phagocytosis, are enhanced by IFN-γ, a cytokine produced in the course of LDV infection,¹⁶  this cytokine was suspected as a mediator of LDV-induced enhancement of antiplatelet antibody pathogenicity. The causative role of IFN-γ was shown in mice genetically unable to respond to this cytokine. The transient effect of LDV on antiplatelet antibody pathogenicity may thus be explained by the peculiar kinetics of immune modulation observed after infection with this virus. Indeed, IFN-γ production after LDV infection reaches a peak 18 hours after virus inoculation, and then rapidly decreases.¹⁶  In addition to IFN-γ, we cannot exclude that other cytokines, including interleukins 2 and 12 and macrophage colony-stimulating factor (M-CSF) may also be involved in this virally induced activation of macrophage functions.

It is not easy to understand how the antigenically different pathogens that have been suspected to trigger childhood acute ITP may lead to autoantibody-mediated thrombocytopenia through antigenic mimicry. However, because macrophage activation and IFN-γ production may be induced by distinct viruses, with diverse kinetics, exacerbation of antibody-mediated thrombocytopenia through enhanced phagocytosis may explain the similar consequences on platelet counts observed after infection of patients with widely different pathogens. This hypothesis is further supported by our observation that MHV, which also triggers IFN-γ production,¹⁷  has an enhancing effect on antiplatelet pathogenicity similar to that found after LDV infection. We can thus hypothesize that the development of childhood acute ITP requires 2 successive events: first, a modification of the potential recognition of platelets by phagocytic cells that might be due to autoantibody production following a common, but still unrecognized cause, and that remains clinically silent, and then a triggering of thrombocytopenia as a consequence of IFN-γ-enhanced phagocytosis by macrophages in the course of infection with one of many different viruses. Therefore, targeting this cytokine-induced macrophage activation might be an alternative for new treatments of autoimmune blood diseases that follow viral infections.

We thank J. Van Snick, P.L. Masson, and B. McIntyre for critical reading of this manuscript; V. Préat and C. Lombry for very kind help in the preparation of liposomes; J-P. Dehoux for immunizing rabbits; M.D. Gonzalez and T. Briet for expert technical assistance; and Roche Diagnostics GmbH, Germany for the gift of clodronate.