Antiidiotype Antibody against Platelet Anti-Gpiiia Contributes to the Regulation of Thrombocytopenia in HIV-1–Itp Patients

Immunological thrombocytopenia is a common complication of HIV-1 infection. The incidence in HIV-1–infected patients of HIV-1–associated immunological thrombocytopenia (HIV-1-ITP) is 0–21% at onset and increases to 30% or more with the development of AIDS 1,2. Kinetic data on platelet survival strongly suggest that early-onset HIV-1–ITP is secondary to increased peripheral destruction of platelets, whereas patients with AIDS are more likely to have decreased platelet production 3. Patients with early-onset HIV-1–ITP have a thrombocytopenic disorder that is clinically indistinguishable from classic autoimmune thrombocytopenia (ATP), seen predominantly in females 4,5,6,7. However, HIV-1–ITP is different from classic ATP with respect to the predominant male incidence and the markedly elevated platelet-associated IgG, IgM, and C3C4, as well as presence of circulating serum immune complexes (CICs) composed of the same 5,6. These complexes contain anti-F(ab′)₂ Abs 8 as well as HIV-1–related Abs 9,10. Affinity purification of IgGs from their CICs with platelets has revealed high-affinity IgG1 Ab against the platelet integrin glycoprotein (GP)IIIa peptide 49–66 11,12. This serum anti-GPIIIa Ab correlates inversely with platelet count (r = 0.71; reference 12) and induces severe thrombocytopenia in mice 12, which can be prevented and/or reversed with GPIIIa 49–66 peptide (reference 12; mouse GPIIIa is 83% homologous with human GPIIIa, and macrophages have Fc receptors for human IgG1).

However, we have recently observed that sera from HIV-1–ITP patients have considerably less anti–GPIIIa 49–66 reactivity compared with ∼150-fold greater reactivity in purified IgG from serum and ∼4,000-fold greater reactivity sequestered in their serum CICs. This suggested the possibility of blocking or antiidiotype Ab against anti-GPIIIa in these patients.

This report documents the presence of blocking IgM antiidiotype antibody (Ab2β and/or Ab2γ) versus anti–GPIIIa 49–66 in these patients, which correlates with their platelet count (r = 0.7, P = 0.001, n = 32) and reverses in vivo induced thrombocytopenia in mice.

The population consists of 37 early-onset HIV-1–infected patients without AIDS (19 homosexuals and 18 drug abusers). 22 were thrombocytopenic, and 15 had normal platelet counts. Five control sera were obtained from healthy laboratory personnel. Seven sera were obtained from classic ATP patients.

IgG was prepared from serum by ion-exchange chromatography 13.

F(ab′)₂ fragments were prepared from purified IgG by pepsin digestion as described 13, and were shown to be free of Fc fragments by SDS-PAGE as well as ELISA 13.

Immune complexes (ICs) were prepared from serum by polyethylene glycol precipitation 5. Precipitates were dissolved in one fifth of their serum volume in 0.01 M PBS, pH 7.4.

IgG and IgM were isolated and purified as described 11. In brief, polyethylene glycol (PEG)-ICs were applied to a staphylococcal protein A affinity column (Sigma-Aldrich). The bound complex was washed with PBS and eluted with 0.1 M glycine buffer, pH 2.5. The eluted material was applied to an acidified sephadex G-200 gel filtration column (Amersham Pharmacia Biotech) preequilibrated with the same elution buffer. Effluents of the IgG peak were isolated, neutralized, dialyzed against PBS, and applied to a rabbit anti-IgM affinity column (ICN Pharmaceuticals, Inc.) prepared from Affi-Gel 10 (Bio-Rad). The flow-through material was free of contaminating IgM by immunoblot and ELISA. Effluents of the IgM peak were isolated, neutralized, dialyzed against PBS, and applied to an anti–Fc receptor affinity column to remove rheumatoid factor. (Fc fragments were prepared by papain digestion 11 and affinity purified on a staphylococcal protein A column; the acid eluate was verified by SDS-PAGE and was coupled to Affi-Gel 10). The flow-through IgM was devoid of rheumatoid factor, as determined by inability to bind to a second Fc column.

Antiplatelet IgG was affinity purified with 10⁸ platelets fixed with 2% paraformaldehyde for 2 h at room temperature, followed by overnight gentle rocking at 4°C, then acid elution and neutralization, as described 11. The IgG subclass, determined by radial immunodiffusion (The Binding Site), was IgG1 with both k and l light chains.

Peptide GPIIIa 49–66, CAPESIEFPVSEARVLED (synthesized by Quality Controlled Biochemicals), was coupled to an affinity column with the heterobifunctional cross-linker sulfo-succinimidyl 4-(N-maleimidomethyl) cyclohexane-1-carboxylate as recommended by the manufacturer (Pierce Chemical Co.; cross-links the resin with NH₂-terminal cysteine of the peptide), and was incubated with 0.4 ml of affinity-purified IgG overnight at 4°C. The column was then washed, eluted at pH 2.5, and neutralized as described 12.

Antibody reactivity was measured by solid-phase ELISA 12,13 using serial doubling dilutions of IgG or IgM on U-shaped polyvinyl microtitre plates (Curtin-Matheson Scientific) preincubated overnight at 4°C with 200 ng of peptide GPIIIa 49–66 or F(ab′)₂ fragment of anti–GPIIIa 49–66 in PBS, and was blocked with 2.5% BSA in PBS. A minimum of two different F(ab′)₂ fragments were used for each experiment. The first Ab, used to detect IgG binding, was a 1:500 dilution of goat F(ab′)₂ anti–human IgG (γ chain specific) coupled to alkaline phosphatase (Sigma-Aldrich). The second Ab, used to detect IgM binding, was a 1:1,000 dilution of goat F(ab′)₂ anti–human IgM (μ chain specific) coupled to alkaline phosphatase (ICN Pharmaceuticals, Inc.). Appropriate enzyme substrate was added, and color was read in an automated micotitre plate reader at 405 nm. In some experiments, bound anti–GPIIIa 49–66 was preincubated with GPIIIa 49–66 peptide for 2 h at room temperature before testing for antiidiotype Ab binding.

IgM Ab titer refers to the reciprocal of the lowest concentration of IgM Ab (μg/well) capable of binding to its antigen, determined by extrapolation of the linear portion of the binding curve to zero binding.

Human affinity-purified anti–GPIIIa 49–66 (25 μg) was injected intraperitoneally into BALB/c mice (Taconic Farms), and blood was drawn from the retroorbital sinus at various times. In some experiments, anti–GPIIIa 49–66 Ab was preincubated with either control IgM or antiidiotype IgM before intraperitoneal injection; in other experiments, control or antiidiotype IgM was given after 4 h of thrombocytopenia. Platelet counts were determined from 20 μl of blood drawn into Unopettes (no. 5855; Becton Dickinson) containing optimal anticoagulant concentration and diluent for quantitating platelet count by phase microscopy.

Fig. 1 demonstrates a comparison of the relative binding reactivity of serum, serum IgG, and purified IC-IgG for peptide GPIIIa 49–66 in a representative experiment of five different patients (Table, HIV-1–ITP). 50% detection sensitivity for the respective Ab cohorts were ∼125, 0.8, and 0.03 μg/well. Thus, serum IgG has ∼150-fold greater reactivity than serum, and IC-IgG has approximately sevenfold greater reactivity than serum IgG (∼4,000-fold greater than serum). Similar studies performed on sera of nonthrombocytopenic HIV-1–infected patients also detected antiplatelet GPIIIa 49–66 in their serum IgG and CIC-IgG, but at considerably lower levels (25- and 35-fold less, respectively; Table; HIV-1 controls). This suggested the possibility of blocking or antiidiotype Ab in serum.

In contradistinction to HIV-1–ITP patients, minimal to absent serum antiplatelet reactivity was noted in seven classic ATP patients, with no enhancement of reactivity noted with serum IgG or IC-IgG (data not shown).

Fig. 2 A demonstrates binding of purified IC-IgG Ab from five different HIV-1–ITP patients to F(ab′)₂ fragments of affinity-purified anti–GPIIIa 49–66. 50% binding was observed at ∼2 μg/ml. Similar results were obtained with a second F(ab′)₂ fragment (data not shown). No binding was obtained with the same five IC-IgG preparations against two different control F(ab′)₂ fragments (one of which is shown in Fig. 2). No binding was obtained with five different control IC-IgG preparations (data not shown).

Fig. 3 A demonstrates poor to partial blocking of binding of IgG antiidiotype Ab to anti–GPIIIa 49–66 with peptide GPIIIa 49–66. Thus, only 20% of Ab binding could be inhibited at a peptide/F(ab′)₂ molar ratio of 1,024:1, and therefore designated blocking (P < 0.05 for last three concentrations of peptide; one-tail Student's t test).

Fig. 2 B demonstrates binding of IgM Ab to anti–GPIIIa 49–66. 50% binding was observed at ∼1.25 μg/ml. Similar results were obtained with a second F(ab′)₂ fragment (data not shown). IgM antiidiotype specificity was next examined against five different control F(ab′)₂ fragments and four different antigens to determine whether this IgM could represent polyclonal germline IgM secreted by CD5⁺ B1 cells 14. No binding was obtained with 16 positively reacting IC-IgM preparations against five different control F(ab′)₂ fragments. One of these experiments is shown in Fig. 2 B. No binding to anti–GPIIIa 49–66 was obtained with five different control IgM preparations made from control subject IC-IgM (Fig. 2 B). No binding was obtained with four different proteins: ovalbumin, soybean trypsin inhibitor, thyroglobulin, or carbonic anhydrase (Fig. 4).

Fig. 3 B demonstrates considerable blocking of binding of IgM antiidiotype Ab to anti–GPIIIa 49–66 with peptide GPIIIa 49–66. Thus, 50% of Ab binding could be inhibited at a peptide/F(ab′)₂ molar ratio of ∼1:6.4, and therefore be designated blocking antibody (Ab2β and/or Ab2γ). No blocking was noted with irrelevant peptide CGYGPKKKRKVGG at a peptide/F(ab′)₂ ratio of 1,024:1. IgM antiidiotype Ab did not bind to peptide GPIIIa 49–66 (five experiments, data not shown).

Fig. 3 C demonstrates blocking of binding of anti–GPIIIa 49–66 to platelets by IgM antiidiotype, not by control IgM or IgG antiidiotype. Thus, 50% of Ab binding to platelets could be inhibited at an IgM/IgG molar ratio of ∼1:10.

The above data suggested that the reason for the relatively weak serum anti–GPIIIa 49–66 reactivity with its Ag was because of the presence of IgM blocking antiidiotype Ab in its serum. We therefore reasoned that if this were true and pathophysiologically relevant, then a positive correlation should be obtained between the antiidiotype titer and the patient's platelet count. This proved to be the case. Fig. 5 demonstrates such a correlation of r = 0.71 (P = 0.001, n = 32). Specific measurements and clinical data are described in Table, lanes 1–9. In these two cohorts, IgM antiidiotype Ab was 12-fold greater in nonthrombocytopenic patients. A similar log mean IgM antiidiotype Ab titer of 1:3,643 was found in 10 additional nonthrombocytopenic HIV-1 patients.

Our previous study demonstrated that human anti–GPIIIa 49–66 Ab could induce significant thrombocytopenia in mice, with nadir at 4 h when injected intraperitoneally (control IgG had no effect; reference 12). We therefore tested the ability of the IgM antiidiotype Ab versus anti–GPIIIa 49–66 to reverse this effect in vivo. Again, this proved to be the case. Thus, Fig. 6 A demonstrates a 70% drop in platelet count induced by 25 μg/ml of anti–GPIIIa 49–66, with reversal to 50–80% of normal by preincubation with IgM antiidiotype/anti–GPIIIa 49–66 ratios of 1:7, respectively. Neither control IgM nor IgG-antiidiotype, at similar ratios, had any effect. To rule out the possibility that the IgM antiidiotype Ab was not operating through increased clearance of the antiplatelet Ab, experiments were also performed after induction of the thrombocytopenia at 4 h. Fig. 6 B demonstrates reversal of thrombocytopenia with the IgM antiidiotype Ab, not with the IgG antiidiotype, thus confirming that the IgM antiidiotype was interfering with the binding of anti-GPIIIa to platelets.

These data clearly indicate the presence of IgG and IgM antiidiotype (Ab2) against anti–GPIIIa 49–66 in HIV-1–ITP patients. Their absence in classic ATP patients suggests that the mechanism of thrombocytopenia is different in both autoimmune disorders. The antiidiotype in HIV-1–ITP patients was both Ab2α as well as blocking (Ab2β and/or Ab2γ) for both isotypes, with blocking IgM Ab2 predominating over IgG Ab2. Several lines of evidence support the conclusion that blocking IgM Ab2 is responsible for the impaired serum reactivity of anti–GPIIIa 49–66: (a) purification of serum IgG and IC-IgG increased anti-GPIIIa reactivity ∼150- and 4,000-fold, respectively; (b) IgM purified from PEG-IC bound to anti–GPIIIa 49–66 in a specific manner (no binding with control IgM); (c) IgM Ab could be blocked from binding to anti–GPIIIa 49–66 with the Ag for anti–GPIIIa 49-66 (not with an irrelevant Ag); (d) anti–GPIIIa 49–66 could be blocked from binding to platelets with IgM antiidiotype, not with control IgM or IgG antiidiotype; and (e) in vivo thrombocytopenia induced in mice with anti–GPIIIa 49–66 could be reversed with purified IC-IgM, not with purified serum IgM or IC-IgG.

These data strongly suggest that the IgM antiidiotype as well as level of anti–GPIIIa 49–66 IgG Ab play a role in regulating early-onset autoimmune HIV-1–ITP in vivo. A correlation of platelet count with IgM antiidiotype of r = 0.7 supports this suggestion, as do our previous observations on the presence of an inverse correlation between anti–GPIIIa 49–66 antibody and platelet count in early-onset HIV-1 infection (thrombocytopenic versus nonthrombocytopenic; reference 12). It should be recognized that multiple factors are likely to regulate the platelet count in HIV-1–ITP patients. These include platelet production, platelet survival, and relative phagocytic function of the reticuloendothelial system. It is therefore not surprising that the correlation coefficient is <1.

The presence of anti–GPIIIa 49–66 Ab in nonthrombocytopenic patients, albeit at 26–35-fold lower reactivity than in HIV-1–ITP patients, is of interest and suggests that low levels of reactivity may be present in most early-onset HIV-1–infected patients. It is possible that this may represent molecular mimicry between anti–HIV-1 Abs and platelet GPIIIa. Indeed, this has been reported for anti–HIV-1–gp120 Ab 15.

Our findings on the presence of blocking IgM antiidiotype in HIV-1–ITP patients is reminiscent of previous observations on IgM Abs blocking natural polyreactive low-affinity Abs in mice as well as humans 16,17,18,19. However, our anti–GPIIIa 49–66 autoantibody preparation is different, in that it is highly specific 11 and contains high-affinity Ab (Kd = 1–2 nM; reference 11). In addition, the purified serum IgG of HIV-1–ITP patients is 150-fold more reactive than serum, compared with the 3–5-fold greater reactivity reported for purified IgG blocking Ab in normal subjects 18. It is possible that anti–GPIIIa 49–66 was originally a polyreactive natural Ab that underwent somatic mutation and selective pressure by antigens (HIV-1), as has been suggested for the development of autoimmune disease 17. This is supported by the presence of lower affinity (Kd = 7–12 nM) Ab as well in our anti–GPIIIa 49–66 preparation 11. The same selective pressure could also apply for natural, polyreactive, low-affinity IgM “antiidiotype” Ab.

Alternatively, the pathogenic potential of low-affinity IgM Ab has recently been demonstrated in a study comparing monoclonal mouse IgM anti-RBC Ab with its IgG class-switch variant 20. In these studies, the RBC binding activity of the IgM Ab was 1,000 times that of its IgG class-switch variant and was related to its pentameric structure, which promoted binding, agglutination of RBCs, and hemolytic anemia. These data indicate that affinity maturation of autoantibodies may not be required for generation of autoantibodies capable of inducing clinical pathology. The same applies for the reactivity of the IgM antiidiotype of our study.

Although the role of antiidiotype Ab in the regulation of the immune response is controversial, a case can be made for its dysregulation in the pathophysiology of some autoimmune diseases: (a) patients with severe, uncontrolled SLE have high levels of anti-DNA Abs and low levels of anti-F(ab′)₂ Abs, whereas patients with quiescent disease have the reverse 21; (b) patients with systemic vasculitis have antiidiotype Abs against antimyeloperoxidase and antineutrophil cytoplasmic antigen, with rise in antiidiotype titer as disease activity subsides 22,23; and (c) a hemophiliac patient with a serious anti–factor VIII (antihemophilic factor) inhibitor developed antiidiotype Ab against anti–factor VIII, which coincided with recovery and the disappearance of the inhibitor 24. A case for antiidiotype dysregulation is indirectly supported by other observations: (a) patients with thyroid autoimmunity have naturally occurring antiidiotype IgM Ab against antimicrosomal Abs 25; (b) patients with myasthenia gravis have naturally occurring antiidiotype Abs against acetylcholine receptor Ab 26; and (c) intravenous γ-globulin infusions containing “antiidiotype” Ab are often effective in the treatment of patients with autoimmune thyroid disease 25, systemic vasculitis 22,23, myasthenia gravis 27, and kawasaki syndrome 28.

Our results contribute to and extend these observations in a more definitive manner in HIV-1–ITP. Our observations demonstrate the presence of high-affinity anti–GPIIIa 49–66 Ab 11,12, specific blocking IgM antiidiotype Ab, a positive correlation between IgM antiidiotype and platelet count, and most importantly, a reversal of in vivo antibody-induced thrombocytopenia with its antiidiotype. These data support the concept that dysregulation of antiidiotype Ab can play a role in the development of autoimmune disease in HIV-1–ITP.

This work was supported by National Institutes of Health grants HL-13336-26 and DA-04315-1A, and by the Dorothy and Seymour Weinstein Platelet Research Fund.