Developing recombinant HPA-1a–specific antibodies with abrogated Fcγ receptor binding for the treatment of fetomaternal alloimmune thrombocytopenia

Fetomaternal alloimmune thrombocytopenia (FMAIT) is caused by maternal generation of antibodies specific for paternal platelet antigens and can lead to fetal intracranial hemorrhage. A SNP in the gene encoding integrin β3 causes a clinically important maternal-paternal antigenic difference; Leu33 generates the human platelet antigen 1a (HPA-1a), whereas Pro33 generates HPA-1b. As a potential treatment to prevent fetal intracranial hemorrhage in HPA-1a alloimmunized pregnancies, we generated an antibody that blocks the binding of maternal HPA-1a–specific antibodies to fetal HPA-1a1b platelets by combining a high-affinity human HPA-1a–specific scFv (B2) with an IgG1 constant region modified to minimize Fcγ receptor–dependent platelet destruction (G1Δnab). B2G1Δnab saturated HPA-1a⁺ platelets and substantially inhibited binding of clinical HPA-1a–specific sera to HPA-1a⁺ platelets. The response of monocytes to B2G1Δnab-sensitized platelets was substantially less than their response to unmodified B2G1, as measured by chemiluminescence. In addition, B2G1Δnab inhibited chemiluminescence induced by B2G1 and HPA-1a–specific sera. In a chimeric mouse model, B2G1 and polyclonal Ig preparations from clinical HPA-1a–specific sera reduced circulating HPA-1a⁺ platelets, concomitant with transient thrombocytopenia. As the Δnab constant region is uninformative in mice, F(ab′)₂ B2G1 was used as a proof of principle blocking antibody and prevented the in vivo platelet destruction seen with B2G1 and polyclonal HPA-1a–specific antibodies. These results provide rationale for human clinical studies.

Fetomaternal alloimmunization to paternal human platelet antigens (HPAs) is the most common cause of severe thrombocytopenia in term neonates (1), with 75% of cases due to alloantibodies against HPA-1a (2–5). One in 4 babies born to HPA-1a–immunized mothers have fewer than 20 × 10⁹ platelets/l (5–10), which leads to intracranial hemorrhage (ICH) in 10% to 20% of all cases with fetomaternal alloimmune thrombocytopenia (FMAIT), from 16 weeks of pregnancy through the postnatal period (8, 11, 12). There is no consensus on the most effective antenatal therapy for FMAIT (13). Intrauterine transfusions (IUTs) of HPA-1a^– platelets are a logistical challenge (14), and the risk of fetal loss is up to 15% (15–18). Several trials have shown the benefit of intravenous immunoglobulin (IVIG) therapy, but in 50% of patients with severe disease (ICH in the previous pregnancy or an initial fetal platelet count of fewer than 20 × 10⁹/l), IVIG therapy does not achieve a safe fetal platelet count (19–23). Although IVIG therapy may reduce the incidence of ICH in this high-risk group without a rise in platelet count (24), ICHs still occur in some cases (21, 25, 26). The most common side effects of IVIG therapy, such as headaches, myalgia, and allergic reactions, can be easily treated (27). However, the infectious risk associated with pooled blood products such as IVIG cannot be dismissed, as a previous outbreak of hepatitis C associated with plasma-derived anti-D immunoglobulin has shown (28, 29), and there is concern over emerging infectious agents such as prions, for which donors are not screened and which are resistant to heat treatment (30). IVIG therapy is expensive and its chronic worldwide shortage well documented (31, 32). Therefore, a safe and effective recombinant alternative for antenatal treatment of FMAIT would be useful.

It has been shown that the binding site for polyclonal HPA-1a antibodies is limited to a finite number of epitopes on the β3 integrin, with Leu33 being a critical residue in the antibody binding site (33). We reasoned that it should be possible to generate a recombinant nondestructive blocking HPA-1a antibody of sufficiently high affinity to block binding of maternal polyclonal HPA-1a antibodies to fetal HPA-1a1b platelets. A potential therapeutic antibody would require an Fc portion to maintain the long half-life of IgG (34) and to mediate placental transport via FcRn (35), removing the need for hazardous intrauterine administration. The precise mechanism of platelet destruction is assumed to involve the high-affinity Fcγ receptors (FcγRs) on effector cells. The therapeutic antibody Fc portion would therefore have to be modified to prevent binding to FcγRs, particularly the high-affinity FcγRI (CD64).

We previously generated a human single-chain variable domain antibody fragment (scFv) of nanomolar affinity (K_d = 6 × 10^–8 M) for HPA-1a from the maternal B cells of a case of FMAIT by phage display (36, 37). The recombinant IgG1 antibody (B2G1) derived from this scFv was sufficiently specific for HPA-1a to permit its use as a routine phenotyping reagent (38).

To generate a complete antibody lacking destructive activity, residues from IgG2 and IgG4 were substituted into regions of the IgG1 CH2 domain involved in binding to FcγRI–FcγRIII and complement C1q. Residues of IgG1 responsible for the immunogenic Gm allotype were replaced by the non-immunogenic residues from IgG2 (Figure 1) (39). Platelets sensitized with the wild-type IgG1 anti–HPA-1a described above have been shown to trigger monocyte chemiluminescence (CL) in an interaction mediated by FcγRI and FcγRII (40). It is therefore logical to hypothesize that the modified recombinant HPA-1a antibody (B2G1Δnab) could block binding of maternal HPA-1a antibodies to platelets, reduce monocyte activation, and ultimately prevent platelet destruction.

Figure 1

Model of human IgG1 indicating the positions of the mutated residues in the modified nondestructive G1Δnab constant region. The immunoglobulin IgG heavy chains are shown in light blue and the immunoglobulin light chains and Fc-associated carbohydrate in dark gray. The red residues were altered by the Δn mutation in order to replace the IgG1 G1m(1,17) allotypic residues Lys214, Asp356, and Leu358 with the corresponding IgG2 residues Thr, Glu, and Met. The blue amino acids are IgG1 residues Ala327, Ala330, and Pro331 changed to the IgG4 residues Gly, Ser, and Ser by the Δa mutation. The green residues of IgG1 (Glu233, Leu234, Leu235, and Gly236) were substituted with the corresponding amino acids of IgG2 (Pro, Val, Ala, and a deleted residue) by the Δb mutation. The image was generated from the PDB file of an IgG1 model (59) using RasMol V2.7.3.

The biallelic HPA-1 system only exists in humans, and there is no satisfactory animal model to investigate the possible use of these recombinant antibodies. Ni and colleagues developed a murine model to look at the efficacy of IVIG therapy, but their study relied on isotypic antibodies generated in β3-deficient (β3^–/– mice) against a β3^+/– fetus (41). We therefore generated β3^–/– mice transplanted with littermate bone marrow transduced with a lentivirus vector containing the human β3 gene ITGB3 encoding either Leu33 (HPA-1a) or Pro33 (HPA-1b). Transplanted mice express a hybrid murine/human αIIbβ3 complex on the platelet surface as previously described (42) but crucially, this complex bears the corresponding HPA-1a or -1b antigen, making the mice suitable for use in studying the effects of natural and recombinant human HPA-1a antibodies on platelet survival in vivo.

This study demonstrates that the modified HPA-1a antibody B2G1Δnab inhibits binding of maternal HPA-1a antibodies from FMAIT cases to platelets and abrogates monocyte CL responses to anti–HPA-1a–coated platelets and that its F(ab′)₂ fragment prevents removal from murine circulation of HPA-1a–expressing platelets by destructive HPA-1a antibodies.

Binding of IgG1 and mutated recombinant HPA-1a antibodies to HPA-1a1b platelets. We assessed the binding characteristics of the parent IgG1 (B2G1) and modified antibody (B2G1Δnab) to the HPA-1a antigen on the surface of HPA-1a1b platelets (matching the fetal phenotype). In flow cytometry and the monoclonal antibody immobilization of platelet antigen (MAIPA) assay, dilutions of B2G1 and B2G1Δnab demonstrated identical levels of binding to HPA-1a1b platelets, with saturation at 5 μg/ml (data not shown). There was no binding to homozygous HPA-1b1b platelets (data not shown).

Competition studies were carried out in the MAIPA using F(ab′)₂ B2G1 as a blocking antibody. Binding of 10 μg/ml B2G1 and B2G1Δnab to platelets was inhibited to the same extent by F(ab′)₂ B2G1, requiring 3, 8, and 50 μg/ml to achieve 50%, 70%, and maximum (98%) inhibition (data not shown).

Monocyte CL responses to IgG1 and mutated recombinant HPA-1a antibodies. We assessed the effect on monocyte activation of the mutations introduced in the constant region of the modified antibody B2G1Δnab. The monocyte CL response to HPA-1a1b platelets sensitized with the parent IgG1 B2G1 reached a maximum at 10 μg/ml (Figure 2, A and B). The CL responses to B2G1Δnab-sensitized HPA-1a1b platelets were reduced to less than 15% of B2G1 values (P < 0.001) across the antibody concentration range (Figure 2, A and B) but were above that observed with the control antibody VAZO-5 or F(ab′)₂ B2G1 (Figure 2, A and B). No CL response was obtained with HPA-1b1b platelets (data not shown).

Figure 2

Monocyte CL response to recombinant HPA-1a antibodies. (A) CL response to HPA-1a1b platelets sensitized with 10 μg/ml of B2G1, B2G1Δnab, F(ab′)₂ B2G1, and a negative control antibody VAZO-5. CL intensity was recorded over 20 cycles of 2.35 minutes each. (B) The CL signal corresponding to the sum of the CL intensity for the 20 cycles (surface under the curve) recorded with decreasing concentrations of B2G1, B2G1Δnab, and VAZO-5. (C) Competition study in the CL assay. HPA-1a1b platelets were sensitized with 10 μg/ml B2G1 and decreasing concentrations of B2G1Δnab. (D) Competition curve generated from the results in C. The CL signal obtained with 10 μg/ml B2G1 in the absence of the competing antibody B2G1Δnab corresponds to 100% residual CL signal on the graph. Data are the mean ± SD of 3 separate experiments.

In competition studies, increasing concentrations of B2G1Δnab inhibited the CL response to 10 μg/ml B2G1 (Figure 2C), requiring 26, 49, and more than 500 μg/ml to achieve 50%, 70%, and maximum (85%) inhibition, respectively (Figure 2D).

Inhibition of platelet binding and CL responses to maternal polyclonal anti–HPA-1a by recombinant HPA-1a antibodies. Assays were performed to determine whether F(ab′)₂ B2G1 could inhibit binding of maternal polyclonal HPA-1a antibodies from 20 FMAIT cases and whether B2G1Δnab could reduce monocyte responses to the same sera. The anti–HPA-1a potencies of the maternal sample ranged from 0.01 to 193 IU/ml, and the monocyte CL signal induced by each sample ranged from 0.3 to 31.2 CL units (Table 1). Each patient was assigned a unique patient number (UPN).

Table 1

Clinical details, HPA-1a antibody potency, CL signal, and inhibition data

For all maternal sera, we established inhibition curves in the MAIPA assay using F(ab′)₂ B2G1 as described above, and the concentrations were calculated for F(ab′)₂ B2G1, which was found to inhibit binding by 50% (50% inhibition of binding [ID50]), 70% (ID70), and by the maximum amount achievable.

For 18 sera, the ID50 values ranged from 4.8 to 616 μg/ml F(ab′)₂ B2G1 and ID70 values from 36 to 2,000 μg/ml F(ab′)₂ B2G1 (Figure 3A). The maximum achievable inhibition of binding ranged from 70% to 97% (Figure 3A), with a F(ab′)₂ B2G1 concentration ranging from 20 to 4,000 μg/ml (Figure 3A). In one case the maternal HPA-1a antibody was too weak to establish an inhibition curve, and in another the F(ab′)₂ B2G1 increased the OD by 3-fold. This particular serum was found by ELISA to contain anti-F(ab′)₂ antibodies (data not shown). In a further experiment, F(ab′)₂ B2G1 was added 20 minutes after pre-sensitization of the platelets with each of 5 maternal sera, at a concentration previously shown to inhibit binding by 50%. The degree of inhibition was comparable with that obtained when both were added concomitantly (Figure 4). This result proves the ability of the recombinant antibodies to displace maternal polyclonal antibodies already bound to platelets. There was, however, no correlation between maternal antibody potency and concentrations of mutant antibody required for inhibition of binding (Table 1).

Figure 3

Competition studies with 20 clinical sera containing HPA-1a antibodies. (A) Concentrations of F(ab′)₂ B2G1 (μg/ml) necessary to inhibit binding in the MAIPA assay by 50% and 70% and to achieve maximum inhibition for each clinical sample. The maximum inhibition achieved (%) is shown in the right panel. In UPN 1, the maternal HPA-1a antibody was too weak to establish an inhibition curve, and in UPN 2 the F(ab′)₂ B2G1 increased platelet binding of maternal HPA-1a antibodies by 3-fold. This particular serum was found by ELISA to contain anti-F(ab′)₂ antibodies. (B) Concentrations of B2G1Δnab (μg/ml) necessary to inhibit CL response by 50% and 70% and to achieve maximum inhibition for each clinical sample. The maximum inhibition achieved (%) is shown in the right panel. In 3 cases (UPNs 1, 10, and 19), the CL signal was too weak to establish an inhibition curve. For the 2 samples containing HPA-5b as well as -1a antibodies (UPNs 8 and 9), maximum inhibition failed to reach 70%.

Figure 4

Competition studies with 5 clinical sera containing HPA-1a antibodies with concomitant and postponed addition of F(ab′)₂ B2G1. To assess the ability of the recombinant antibodies to displace polyclonal antibodies already bound to platelets, F(ab′)₂ B2G1 was added 20 minutes after pre-sensitization of the platelets with each of 5 maternal sera, at a concentration previously shown to inhibit binding by 50% and inhibition compared with that obtained when both were added concomitantly.

Inhibition curves were generated in the CL assay for all clinical samples using increasing concentrations of B2G1Δnab mixed with a fixed amount of polyclonal serum. The concentrations of B2G1Δnab that inhibited the CL response to each polyclonal serum by 50%, 70%, and the maximum amount achievable were calculated. In 17 cases, the ID50 values ranged from 6 to 560 μg/ml B2G1Δnab and the ID70 values from 10 to 1,940 μg/ml B2G1Δnab (Figure 3B). The maximum inhibition achieved in the CL assay ranged from 72% to 95% for the 15 pure anti–HPA-1a sera (Figure 3B, right hand-side panel), requiring 6–3,200 μg/ml B2G1Δnab (Figure 3B). For 2 sera known to contain a mixture of HPA-1a and -5b antibodies, maximum inhibition was 55% and 65%, respectively (the platelets used in the assay were from a HPA-5a5b heterozygous donor). In 3 cases the CL signal was too weak to establish an inhibition curve. There was no correlation between concentrations of inhibitory antibody required in the CL studies and maternal antibody potency, nor between concentrations required for the same degree of inhibition in MAIPA and CL.

A murine model to study inhibition of HPA-1a platelet destruction by blocking antibodies in vivo. Bone marrow from β3^–/– mice was transduced with a viral construct containing the human ITGB3 cDNA encoding either leucine or proline at position 33 (HPA-1a or -1b, respectively), and the transduced bone marrow was transplanted into lethally irradiated β3^–/– mice. Upon bone marrow recovery, the transplanted mice had a chimeric platelet population, containing platelets generated from the original β3^–/– megakaryocytes and platelets expressing the human β3 (Huβ3 Pt) and the corresponding HPA-1a or -1b epitope (Figure 5). Huβ3 Pt chimerism levels (the fraction of Huβ3 Pt present in circulation expressed as a percentage of the total platelet population) varied from 10% to 60% in the HPA-1a mice (n = 10) and from 54% to 81% in the HPA-1b mice (n = 4). Specific binding of FITC-B2G1 confirmed the expression of the HPA-1a antigen on platelets circulating within HPA-1a mice. HPA-1a was not detected on the platelets of HPA-1b or β3^–/– mice (Figure 5).

Figure 5

Platelet chimerism in β3^–/– transplanted mice and specific expression of the HPA-1a epitope. Lethally irradiated β3^–/– mice were transplanted with bone marrow from β3^–/– mice transduced with a viral construct containing the human ITGB3 cDNA encoding either leucine or proline at position 33 (HPA-1a or -1b, respectively). Upon bone marrow recovery, the transplanted mice had a chimeric platelet population, containing platelets generated from the original β3^–/– megakaryocytes and Huβ3 Pt as shown in the top panel, where platelets were stained with the PE-labeled human β3–specific antibody VI-PL2. The bottom panel shows specific binding of the FITC-labeled HPA-1a recombinant antibody B2G1 to platelets in mice transplanted with bone marrow transduced with the Leu33 human β3.

We first assessed the in vivo destructive effect of the recombinant IgG1 HPA-1a antibody B2G1 on the Huβ3 Pt by i.p. injection. To be able to compare the different animals entered into the experiment, the remaining fraction of Huβ3 Pt present in circulation after i.p. injection was expressed as a percentage of the baseline Huβ3 Pt chimerism for each mouse. In calculating the amount of antibody necessary to achieve thrombocytopenia, we assumed complete absorption from the i.p. injection into a 2-ml circulatory blood volume. i.p. injection of 25 μg B2G1 (i.e., ~12.5 μg/ml in circulation) in 3 HPA-1a mice (baseline Huβ3 Pt chimerisms of 49%, 21%, and 12%) resulted in a drop in Huβ3 Pt levels of 18%–67% at 1 hour, reaching maximum reduction of 67%–93% at 4 hours (Figure 6A). In the 2 highest expressors, this was reflected by a decrease in platelet count proportional to their baseline Huβ3 Pt chimerism (Figure 6B). There was no effect of B2G1 on Huβ3 Pt or platelet count in HPA-1b (Figure 6, A and B) or in β3^–/– mice (Figure 6B). Huβ3 Pt levels returned to baseline in all HPA-1a mice 7–10 days after the experiment.

Figure 6

Platelet destruction studies with HPA-1a recombinant antibodies in the mice. β3^–/– mice were transplanted with murine bone marrow transduced with the human β3 integrin encoding with either Leu33 (HPA-1a) or Pro33 (HPA-1b). (A) The proportion of Huβ3 Pt in 3 HPA-1a mice (solid lines) and 2 HPA-1b mice (dotted lines) after i.p. injection of 25 μg B2G1. Huβ3 Pt chimerism for each mouse at time 0 (49%, 21%, and 12%) corresponds to 100% Huβ3 Pt on the graph. (B) Platelet counts in the 3 HPA-1a (solid lines), 2 HPA-1b (dotted lines), and 3 β3^–/– (gray lines) mice after 25 μg B2G1 i.p. (C) The proportion of Huβ3 Pt in 3 HPA-1a mice (baseline Huβ3 Pt chimerism of 53%, 21%, and 10%) after i.p. injection of 25 μg B2G1 alone (solid lines) and after 25 μg B2G1 plus i.p. boluses of 150 μg F(ab′)₂ B2G1 (dotted lines) at times shown.

Since the mouse FcγR profile is different from that in humans (43), using the B2G1Δnab construct whose constant region has been tailored to abrogate binding to human FcγR would be uninformative. Therefore, we opted to use a F(ab′)₂ construct of B2G1 that lacks a constant region altogether and therefore does not bind to either human or mouse FcγR as a proof of principle protective antibody. To assess the ability of a F(ab′)₂ fragment of B2G1 to inhibit B2G1-induced platelet clearance, competition studies were performed in 3 HPA-1a mice (Huβ3 Pt chimerism of 53%, 21%, and 10%). Platelet destruction studies were first performed as described above using 25 μg B2G1 i.p. injection, and the clearance of Huβ3 Pt was confirmed (Figure 6C, solid lines). After the peripheral platelet counts returned to normal, the mice were entered into the competition study. Given the shorter half-life of a F(ab′)₂ fragment compared with a full-length antibody in vivo, the mice had a loading dose of 150 μg F(ab′)₂ B2G1 i.p. 1 hour prior to injecting 25 μg B2G1 i.p., followed by 150 μg F(ab′)₂ B2G1 i.p. at 0, 1, and 3 hours (Figure 6C). In all 3 mice, the proportion of Huβ3 Pt remaining in circulation after injection of B2G1 in the presence of F(ab′)₂ B2G1 was close to baseline values, supporting its protective effect (Figure 6C, dotted lines).

To carry out equivalent in vivo studies using polyclonal FMAIT sera, we generated 2 polyclonal IgG preparations (CAM00D and CAM00E) from 2 FMAIT cases. CAM00D and CAM00E anti–HPA-1a potency was measured by MAIPA as 2,500 and 70 IU/ml, respectively (high concentrations were deliberately generated to take into account the dilution in the mice’s circulatory volumes). Competition studies of binding were carried out using MAIPA with F(ab′)₂ B2G1 as described above, using dilutions of 1/200 (12.5 IU/ml) and 1/40 (1.75 IU/ml) for CAM00D and CAM00E, respectively. Results corrected for the dilution factor are shown in Figure 7A. ID50 and ID70 were reached at 5 and 20 mg F(ab′)₂ B2G1, respectively, for CAM00D and at 0.4 and 2.4 mg F(ab′)₂ B2G1, respectively, for CAM00E. Maximum inhibition was 82% for CAM00D with 200 mg F(ab′)₂ B2G1 and 83% for CAM00E with 20 mg F(ab′)₂ B2G1. These high values reflect the high concentration of HPA-1a antibodies in these 2 preparations.

Figure 7

Platelet destruction studies in the murine model with polyclonal HPA-1a antibodies and F(ab′)₂ B2G1. (A) Binding of the human polyclonal Ig anti–HPA-1a preparations CAM00D and CAM00E to HPA-1a heterozygous platelets by MAIPA in the presence of increasing concentrations of the F(ab′)₂ fragment of B2G1. (B) The proportion of Huβ3 Pt in 3 HPA-1a mice (baseline Huβ3 Pt chimerism of 49%, 60%, and 34%) after i.p. injection of CAM00D alone (100 μl for the first 2 mice and 10 μl for the last one) (solid lines) and the proportion of Huβ3 Pt in 3 HPA-1a mice (baseline Huβ3 Pt chimerism of 67%, 83%, and 22%) after CAM00D (100 μl for the first 2 mice; 10 μl for the last one) plus i.p. boluses of F(ab′)₂ B2G1 (260, 130, 130, and 130 μg for the first 2 mice; 50, 25, 12.5, and 12.5 μg for the last one) (dotted lines) at the times indicated. (C) The proportion of Huβ3 Pt in 2 HPA-1a mice (baseline Huβ3 Pt chimerism of 26% and 21%) after i.p. injection of 200 μl CAM00E alone (solid lines) and after CAM00E plus i.p. boluses of 100, 50, 25, and 25 μg F(ab′)₂ B2G1 (dotted lines) at times shown.

Two mice (Huβ3 Pt chimerism of 49% and 60%, respectively) were injected i.p. with 100 μl CAM00D (i.e., ~125 IU/ml in circulation) and 1 mouse (Huβ3 Pt chimerism of 34%) with 10 μl (i.e. ~12.5 IU/ml in circulation). In all 3 cases, significant platelet destruction was observed, with Huβ3 Pt falling to 9%–17% of baseline after 4 hours (Figure 7B). No effect was seen in HPA-1b or β3^–/– mice (data not shown). Competition studies were carried out in 3 mice with Huβ3 Pt chimerism equivalent to that of the mice used in the destructive studies (Huβ3 Pt chimerism of 67%, 83%, and 22%, respectively). The first 2 received 100 μl CAM00D and F(ab′)₂ B2G1 boluses of 260, 130, 130, and 130 μg at –1, 0, 1, and 3 hours, and the last mouse received 10 μl CAM00D and F(ab′)₂ B2G1 boluses of 50, 25, 12.5, and 12.5 μg at the same times. In all 3 cases, platelet destruction was prevented by the competing antibody (Figure 7B).

Two mice (Huβ3 Pt chimerism of 26% and 21%) were injected i.p. with 200 μl CAM00E (~7 IU/ml in circulation). In both mice, platelet destruction was observed, with Huβ3 Pt falling to 45%–55% of baseline after 4 hours (Figure 7C). No effect was seen in HPA-1b or β3^–/– mice (data not shown). Competition studies were carried out in the same mice with F(ab′)₂ B2G1 boluses of 100, 50, 25, and 25 μg given at –1, 0, 1, and 3 hours. In both mice, platelet destruction was completely prevented by the competing antibody (Figure 7C).

The studies reported here confirm that it is possible to inhibit the binding of human polyclonal HPA-1a antibodies to HPA-1a1b platelets with a single human recombinant HPA-1a antibody (B2G1), and that the addition of a modified constant region (B2G1Δnab) abrogates monocyte CL responses to anti–HPA-1a sensitized platelets. Furthermore, we show in a unique murine model that HPA-1a antibody-induced platelet clearance is prevented in vivo by F(ab′)₂ B2G1.

In competition studies, 70%–95% inhibition of anti–HPA-1a binding to platelets was seen with all but one clinical sera tested. Anti- F(ab′)₂ antibodies were later found in this sample (data not shown), which may explain this discrepant result. The concentration of F(ab′)₂ B2G1 necessary to inhibit platelet binding was variable but showed no correlation with maternal anti–HPA-1a potency. This may be because the potency of each polyclonal serum depends on both the concentration and affinity of each of its constituent antibody clones with variation in on/off rates for antigen binding. There were sera for which complete inhibition of binding could be achieved at low concentrations of inhibitor antibody (800 μg/ml) (UPNs 3 and 6), and in contrast, there were also sera that showed more than 20% residual binding (UPNs 4, 10, 12, 13, and 14), even with 4,000 μg/ml of F(ab′)₂ B2G1 (Table 1). This cannot be explained by major differences in affinity between B2G1 (K_d = 6 × 10^–8 M) (37) and the polyclonal sera, as there is good evidence that even high-affinity polyclonal antibodies have at most a K_d of 0.3 × 10^–9 to 1 × 10^–9 M (44). HPA-1a polyclonal sera have different footprints on the β3 integrin (45, 46): type I antibodies, which bind to the plexin semaphorin integrin (PSI) domain (e.g., the first 54 residues of β3), and type II antibodies, which require residues outside the PSI domain. The competition results may therefore also reflect the differences between the antibody footprints covered by the maternal polyclonal antibodies and that covered by our monoclonal antibody (which belongs to the type II category).

Monocyte CL responses to platelets sensitized with the modified HPA-1a antibody were reduced by 85% when compared with the IgG1 parent antibody. These results are in keeping with the previous observation that the Δnab mutations reduced the binding of a human RhD IgG1Δnab antibody to FcγRI by 10⁴-fold (39) and to the stimulatory receptor FcγRIIa by 10-fold (47). The mutated RhD antibodies, when coated onto RhD-positive red cells, also failed to stimulate monocyte CL responses (39), a parameter previously shown to correlate with the severity of hemolytic disease of the newborn (48). However, in contrast to the complete absence of monocyte CL responses observed in the red cell studies, B2G1Δnab elicited a residual response in excess of that seen with a F(ab′)₂ anti–HPA-1a antibody that lacks a constant region altogether. This difference may be an in vitro artefact explained by P-selectin–mediated direct adhesion of platelets to monocytes, which enhances both the rate and, at low antibody concentrations, the magnitude of the monocyte CL response to anti–HPA-1a–sensitized platelets (40).

In keeping with the effect on platelet binding, we have shown that monocyte CL responses induced by heterozygous platelets sensitized with maternal HPA-1a antibodies could be inhibited by more than 70% in all cases in which anti–HPA-1a was present alone. For the 2 sera that also contained anti–HPA-5b, maximum inhibition was notably less (55% and 65%), as the platelets used in the experiment were from a heterozygous HPA-5a5b donor. Since the epitopes defining the HPA-5 antigens are on platelet glycoprotein Ia (α2 integrin), these findings are consistent with the inhibitory effect of the B2 mutated monoclonal antibodies being blockade of the interaction between maternal anti–HPA-1a and its epitope.

The murine studies confirmed the effectiveness of using a blocking HPA-1a antibody with reduced binding to FcγR to prevent platelet destruction. The B2G1Δnab construct in mice competition studies would have been uninformative because human IgG2 and IgG4 antibodies from which the mutations were derived would interact efficiently with murine FcγR (43, 49). Instead, as proof of principle, we opted to use a F(ab′)₂ construct that lacks an Fc domain altogether and therefore does not bind to either human or mouse FcRs. Having demonstrated that the recombinant antibody B2G1 and 2 anti–HPA-1a polyclonal IgG preparations caused clearance only of platelets that expressed the HPA-1a alloantigen, we were able to show that administration of the protective competing antibody F(ab′)₂ B2G1 prevented platelet destruction in all 8 mice studied. Interestingly, the dose of F(ab′)₂ B2G1 necessary to achieve this correlated with the polyclonal anti–HPA-1a potencies and was consistent with the results obtained in the in vitro competition studies.

These results would support studies in human subjects to address safety, efficacy, and dose. In terms of safety, we have already shown that the B2 series of antibodies have no significant adverse effects on platelet activation and function (50). HPA-1a antibodies also bind to endothelial cells αvβ3, but there is published evidence that B2G1 does not alter endothelial cell growth and activation status (51). Moreover, we have shown that ligand binding to αvβ3 is unaffected in the presence of B2G1 (C. Ghevaert, unpublished observations). Finally, we have minimized the potential immunogenicity by the use of a fully human antibody and removal of the Gm allotypic residues.

With regard to efficacy, we have previously demonstrated improved intravascular survival in RhD-positive healthy volunteers of autologous red cells sensitized with an RhD antibody containing the same Δnab constant region compared with red cells sensitized with the parent IgG1 RhD antibody (52), showing that the encouraging results demonstrated in vitro in the CL assay translate to an in vivo protective effect in humans. Whether these results also apply to platelets sensitized with the modified anti–HPA-1a will be assessed in a forthcoming human volunteer study.

With regard to dosing, there is good evidence that in cases affected with anti–HPA-1a antibodies, ICH tends to occur only in fetuses with a platelet concentration of less than 20 × 10⁹/l (18). Therefore, rather than restoring a normal platelet count, our aim with antenatal therapy should be to bring the fetal count into the absolutely safe zone of more than 50 × 10⁹/l. In hemolytic disease of the newborn, clinically significant hemolysis is associated with CL responses of more than 30% of the maximum response obtained with a control antibody (48). With HPA-1a antibodies, there appears to be no correlation between either in vitro potency or monocyte CL responses to maternal anti–HPA-1a and fetal platelet count (53). Therefore, at this stage it would be difficult to estimate by how much maternal antibodies would have to be blocked to generate a clinically significant effect. This particular question will have to be answered in clinical trials, where fetal blood sampling will be necessary to assess dose response. It is, however, possible to calculate the concentration of therapeutic antibody required to achieve a given degree of inhibition for each patient from the results of an in vitro competition assay, as demonstrated in the murine study. Whether the 15% intrinsic activity of B2G1Δnab will limit clinical efficacy is also unknown.

It is conceivable that the recombinant antibody could initially be assessed in patients undergoing therapy with serial IUTs, in whom the antibody could be administered straight into the fetal circulation, along with a platelet transfusion. The effect on the fetal platelet count and the requirement for further IUTs could then be assessed by serial fetal blood sampling in a similar fashion to current ongoing IVIG trials (26). The long-term aim, however, is to remove the risk of cordocentesis altogether by administering the antibody to the mother. We have already demonstrated that our Fc modifications do not abrogate placental transport (54) and binding to the placental transport receptor FcRn is unaffected.

Our in vitro data give some clues as to what concentration and total dose of antibody might be required for maternal administration. Assuming an adult plasma volume of 3 l and equilibrium between maternal and fetal plasma, to reach a concentration of antibody capable of inhibiting 70% of binding and CL response as was achieved here would require between 30 mg and 6 g of recombinant antibody. Such doses can be generated by recombinant technology, as shown by the dose of monoclonal anti-CD20 (rituximab) used for lymphoma therapy (55). Depending on the results of pharmacokinetic studies, multiple boluses of antibody may have to be administered repeatedly to protect from ICH from 16 weeks gestation until delivery.

In conclusion, this study has shown for what we believe to be the first time that it is possible to prevent binding of polyclonal sera to the HPA-1a epitope using a single recombinant high-affinity human antibody. We have shown that the modifications introduced in the constant region aimed at reducing binding to FcγRs substantially decrease monocyte response to platelets sensitized with the modified antibody and, furthermore, that monocyte responses to polyclonal HPA-1a antibodies can be abrogated by the nondestructive antibody. Finally, using a unique murine model, we have shown in proof of principle studies that in vivo platelet destruction by polyclonal anti–HPA-1a can be prevented by a blocking antibody lacking a destructive constant region. Although the program of clinical evaluations will require extensive discussion with obstetricians and regulators, these results would support progression to human studies.

We are grateful to Rosey Mushens (IBGRL) for cell culture and antibody purification and to the Platelet Immunology Reference Laboratory at the NHS Blood and Transplant for their help with the MAIPA assay. This work was supported by grants from the National Blood Service (to C. Ghevaert) and the NIH (HL68138, to D.A. Wilcox). The work was also supported by award R0168138 from the National Heart, Lung, and Blood Institute of the NIH. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Heart, Lung, and Blood Institute.

Address correspondence to: Cedric Ghevaert, Division of Transfusion Medicine, NHS Blood and Transplant, Long Road, Cambridge CB2 2PT, United Kingdom. Phone: 44-7712-179785; Fax: 44-1223-548136; E-mail: cedric.ghevaert@nhsbt.nhs.uk.

Nonstandard abbreviations used: CL, chemiluminescence; FcγR, Fcγ receptor; FMAIT, fetomaternal alloimmune thrombocytopenia; HPA-1a, human platelet antigen 1a; Huβ3 Pt, platelets expressing human β3; IBGRL, International Blood Group Reference Laboratory; ICH, intracranial hemorrhage; ID50, 50% inhibition of binding; IUT, intrauterine transfusion; IVIG, intravenous immunoglobulin; MAIPA, monoclonal antibody immobilization of platelet antigen (assay); UPN, unique patient number.

Conflict of interest: K.L. Armour, M.R. Clark, and L.M. Williamson have filed a patent application for the use of the mutant IgG constant region studied here.

Reference information: J. Clin. Invest.118:2929–2938 (2008). doi:10.1172/JCI34708

See the related articles at Helicobacter pylori eradication shifts monocyte Fcγ receptor balance toward inhibitory FcγRIIB in immune thrombocytopenic purpura patients, Fc receptors in immune thrombocytopenias: a target for immunomodulation?, and Helicobacter pylori eradication shifts monocyte Fcγ receptor balance toward inhibitory FcγRIIB in immune thrombocytopenic purpura patients.

1. Burrows, R.F., Kelton, J.G. 1993. Fetal thrombocytopenia and its relation to maternal thrombocytopenia. N. Engl. J. Med. 329:1463-1466.
   View this article via: CrossRef PubMed Google Scholar
2. Mueller-Eckhardt, C., et al. 1989. 348 cases of suspected neonatal alloimmune thrombocytopenia. Lancet. 1:363-366.
   View this article via: PubMed Google Scholar
3. Berry, J.E., et al. 2000. Detection of Gov system antibodies by MAIPA reveals an immunogenicity similar to the HPA-5 alloantigens. Br. J. Haematol. 110:735-742.
   View this article via: CrossRef PubMed Google Scholar
4. Davoren, A., Curtis, B.R., Aster, R.H., McFarland, J.G. 2004. Human platelet antigen-specific alloantibodies implicated in 1162 cases of neonatal alloimmune thrombocytopenia. Transfusion. 44:1220-1225.
   View this article via: CrossRef PubMed Google Scholar
5. Williamson, L.M., et al. 1998. The natural history of fetomaternal alloimmunization to the platelet-specific antigen HPA-1a (PlA1, Zwa) as determined by antenatal screening. Blood. 92:2280-2287.
   View this article via: PubMed Google Scholar
6. Jaegtvik, S., et al. 2000. Neonatal alloimmune thrombocytopenia due to anti-HPA 1a antibodies; the level of maternal antibodies predicts the severity of thrombocytopenia in the newborn. BJOG. 107:691-694.
   View this article via: CrossRef PubMed Google Scholar
7. Durand-Zaleski, I., et al. 1996. Screening primiparous women and newborns for fetal/neonatal alloimmune thrombocytopenia: a prospective comparison of effectiveness and costs. Immune Thrombocytopenia Working Group. Am. J. Perinatol. 13:423-431.
   View this article via: PubMed Google Scholar
8. Dreyfus, M., et al. 1997. Frequency of immune thrombocytopenia in newborns: a prospective study. Immune Thrombocytopenia Working Group. Blood. 89:4402-4406.
   View this article via: PubMed Google Scholar
9. Turner, M.L., et al. 2005. Prospective epidemiologic study of the outcome and cost-effectiveness of antenatal screening to detect neonatal alloimmune thrombocytopenia due to anti–HPA-1a. Transfusion. 45:1945-1956.
   View this article via: CrossRef PubMed Google Scholar
10. Kjeldsen-Kragh, J., et al. 2007. A screening and intervention program aimed to reduce mortality and serious morbidity associated with severe neonatal alloimmune thrombocytopenia. Blood. 110:833-839.
    View this article via: CrossRef PubMed Google Scholar
11. Giovangrandi, Y., et al. 1990. Very early intracranial haemorrhage in alloimmune fetal thrombocytopenia. Lancet. 336:310.
    View this article via: PubMed Google Scholar
12. Murphy, M.F., et al. 1994. Antenatal management of fetomaternal alloimmune thrombocytopenia--report of 15 affected pregnancies. Transfus. Med. 4:281-292.
    View this article via: CrossRef PubMed Google Scholar
13. Engelfriet, C.P., et al. 2003. Prenatal management of alloimmune thrombocytopenia of the fetus. Vox. Sang. 84:142-149.
    View this article via: CrossRef PubMed Google Scholar
14. Ranasinghe, E., et al. 2001. Provision of platelet support for fetuses and neonates affected by severe fetomaternal alloimmune thrombocytopenia. Br. J. Haematol. 113:40-42.
    View this article via: CrossRef PubMed Google Scholar
15. Overton, T.G., Duncan, K.R., Jolly, M., Letsky, E., Fisk, N.M. 2002. Serial aggressive platelet transfusion for fetal alloimmune thrombocytopenia: platelet dynamics and perinatal outcome. Am. J. Obstet. Gynecol. 186:826-831.
    View this article via: CrossRef PubMed Google Scholar
16. Buscaglia, M., et al. 1996. Percutaneous umbilical blood sampling: indication changes and procedure loss rate in a nine years’ experience. Fetal Diagn.Ther. 11:106-113.
    View this article via: PubMed Google Scholar
17. Paidas, M.J., et al. 1995. Alloimmune thrombocytopenia: fetal and neonatal losses related to cordocentesis. Am. J. Obstet. Gynecol. 172:475-479.
    View this article via: CrossRef PubMed Google Scholar
18. Ghevaert, C., et al. 2007. Management and outcome of 200 cases of fetomaternal alloimmune thrombocytopenia. Transfusion. 47:901-910.
    View this article via: CrossRef PubMed Google Scholar
19. Bussel, J.B., et al. 1996. Antenatal management of alloimmune thrombocytopenia with intravenous gamma-globulin: a randomized trial of the addition of low-dose steroid to intravenous gamma-globulin. Am. J. Obstet. Gynecol. 174:1414-1423.
    View this article via: CrossRef PubMed Google Scholar
20. Berkowitz, R.L., et al. 2006. Parallel randomized trials of risk-based therapy for fetal alloimmune thrombocytopenia. Obstet. Gynecol. 107:91-96.
    View this article via: PubMed Google Scholar
21. Kaplan, C., Murphy, M.F., Kroll, H., Waters, A.H. 1998. Feto-maternal alloimmune thrombocytopenia: antenatal therapy with IvIgG and steroids--more questions than answers. European Working Group on FMAIT. Br. J. Haematol. 100:62-65.
    View this article via: CrossRef PubMed Google Scholar
22. Birchall, J.E., Murphy, M.F., Kaplan, C., Kroll, H. 2003. European collaborative study of the antenatal management of feto-maternal alloimmune thrombocytopenia. Br. J. Haematol. 122:275-288.
    View this article via: CrossRef PubMed Google Scholar
23. Radder, C.M., Brand, A., Kanhai, H.H. 2001. A less invasive treatment strategy to prevent intracranial hemorrhage in fetal and neonatal alloimmune thrombocytopenia. Am. J. Obstet. Gynecol. 185:683-688.
    View this article via: CrossRef PubMed Google Scholar
24. Kanhai, H.H., van den Akker, E.S., Walther, F.J., Brand, A. 2006. Intravenous immunoglobulins without initial and follow-up cordocentesis in alloimmune fetal and neonatal thrombocytopenia at high risk for intracranial hemorrhage. Fetal Diagn.Ther. 21:55-60.
    View this article via: CrossRef PubMed Google Scholar
25. Kroll, H., et al. 1994. Maternal intravenous immunoglobulin treatment does not prevent intracranial haemorrhage in fetal alloimmune thrombocytopenia. Transfus. Med. 4:293-296.
    View this article via: CrossRef PubMed Google Scholar
26. Berkowitz, R.L., Bussel, J.B., McFarland, J.G. 2006. Alloimmune thrombocytopenia: state of the art 2006. Am. J. Obstet. Gynecol. 195:907-913.
    View this article via: CrossRef PubMed Google Scholar
27. El Shanawany, T., Jolles, S. 2007. Intravenous immunoglobulin and autoimmune disease. Ann. N. Y. Acad. Sci. 1110:507-515.
    View this article via: CrossRef PubMed Google Scholar
28. Power, J.P., et al. 1995. Hepatitis C infection from anti-D immunoglobulin. Lancet. 346:372-373.
    View this article via: CrossRef PubMed Google Scholar
29. Smith, D.B., et al. 1999. A second outbreak of hepatitis C virus infection from anti-D immunoglobulin in Ireland. Vox Sang. 76:175-180.
    View this article via: CrossRef PubMed Google Scholar
30. Ironside, J.W. 2006. Variant Creutzfeldt-Jakob disease: risk of transmission by blood transfusion and blood therapies. Haemophilia. 12(Suppl. 1):8-15.
    View this article via: PubMed Google Scholar
31. Pendergrast, J.M., Sher, G.D., Callum, J.L. 2005. Changes in intravenous immunoglobulin prescribing patterns during a period of severe product shortages, 1995–2000. Vox Sang. 89:150-160.
    View this article via: CrossRef PubMed Google Scholar
32. Bayry, J., Kazatchkine, M.D., Kaveri, S.V. 2007. Shortage of human intravenous immunoglobulin — reasons and possible solutions. Nat. Clin. Pract. Neurol. 3:120-121.
    View this article via: PubMed Google Scholar
33. Goldberger, A., Kolodziej, M., Poncz, M., Bennett, J.S., Newman, P.J. 1991. Effect of single amino acid substitutions on the formation of the PlA and Bak alloantigenic epitopes. Blood. 78:681-687.
    View this article via: PubMed Google Scholar
34. Greenwood, J., and Clark, M.R. 1993. Protein engineering of antibody molecules for prophylactic and therapeutic applications in man. Academic Titles. Nottingham, United Kingdom. 85–100.
35. Firan, M., et al. 2001. The MHC class I-related receptor, FcRn, plays an essential role in the maternofetal transfer of gamma-globulin in humans. Int. Immunol. 13:993-1002.
    View this article via: CrossRef PubMed Google Scholar
36. Griffin, H.M., Ouwehand, W.H. 1995. A human monoclonal antibody specific for the leucine-33 (P1A1, HPA-1a) form of platelet glycoprotein IIIa from a V gene phage display library. Blood. 86:4430-4436.
    View this article via: PubMed Google Scholar
37. Santoso, S., et al. 2006. A naturally occurring LeuVal mutation in beta3-integrin impairs the HPA-1a epitope: the third allele of HPA-1. Transfusion. 46:790-799.
    View this article via: CrossRef PubMed Google Scholar
38. Garner, S.F., et al. 2000. A rapid one-stage whole-blood HPA-1a phenotyping assay using a recombinant monoclonal IgG1 anti–HPA-1a. Br. J. Haematol. 108:440-447.
    View this article via: CrossRef PubMed Google Scholar
39. Armour, K.L., Clark, M.R., Hadley, A.G., Williamson, L.M. 1999. Recombinant human IgG molecules lacking Fcgamma receptor I binding and monocyte triggering activities. Eur. J. Immunol. 29:2613-2624.
    View this article via: CrossRef PubMed Google Scholar
40. Turner, C.P., Hadley, A.G. 2003. The role of P-selectin in the immune destruction of platelets. Br. J. Haematol. 121:623-631.
    View this article via: CrossRef PubMed Google Scholar
41. Ni, H., et al. 2006. A novel murine model of fetal and neonatal alloimmune thrombocytopenia: response to intravenous IgG therapy. Blood. 107:2976-2983.
    View this article via: CrossRef PubMed Google Scholar
42. Fang, J., et al. 2005. Therapeutic expression of the platelet-specific integrin, alphaIIbbeta3, in a murine model for Glanzmann thrombasthenia. Blood. 106:2671-2679.
    View this article via: CrossRef PubMed Google Scholar
43. Mestas, J., Hughes, C.C. 2004. Of mice and not men: differences between mouse and human immunology. J. Immunol. 172:2731-2738.
    View this article via: PubMed Google Scholar
44. Bye, J.M., et al. 1992. Germline variable region gene segment derivation of human monoclonal anti-Rh(D) antibodies. Evidence for affinity maturation by somatic hypermutation and repertoire shift. J. Clin. Invest. 90:2481-2490.
    View this article via: JCI CrossRef PubMed Google Scholar
45. Valentin, N., Visentin, G.P., Newman, P.J. 1995. Involvement of the cysteine-rich domain of glycoprotein IIIa in the expression of the human platelet alloantigen, PlA1: evidence for heterogeneity in the humoral response. Blood. 85:3028-3033.
    View this article via: PubMed Google Scholar
46. Stafford, P., et al. 2008. Immunologic and structural analysis of eight novel domain-deletion beta3 integrin peptides designed for detection of HPA-1 antibodies. J. Thromb. Haemost. 6:366-375.
    View this article via: CrossRef PubMed Google Scholar
47. Armour, K.L., van de Winkel, J.G., Williamson, L.M., Clark, M.R. 2003. Differential binding to human FcgammaRIIa and FcgammaRIIb receptors by human IgG wildtype and mutant antibodies. Mol. Immunol. 40:585-593.
    View this article via: CrossRef PubMed Google Scholar
48. Hadley, A.G., et al. 1998. The ability of the chemiluminescence test to predict clinical outcome and the necessity for amniocenteses in pregnancies at risk of haemolytic disease of the newborn. Br. J. Obstet.Gynaecol. 105:231-234.
    View this article via: PubMed Google Scholar
49. Isaacs, J.D., Clark, M.R., Greenwood, J., Waldmann, H. 1992. Therapy with monoclonal antibodies. An in vivo model for the assessment of therapeutic potential. J. Immunol. 148:3062-3071.
    View this article via: PubMed Google Scholar
50. Joutsi-Korhonen, L., et al. 2004. The effect of recombinant IgG antibodies against the leucine-33 form of the platelet beta3 integrin (HPA-1a) on platelet function. Thromb. Haemost. 91:743-754.
    View this article via: PubMed Google Scholar
51. Radder, C.M., Beekhuizen, H., Kanhai, H.H., Brand, A. 2004. Effect of maternal anti–HPA-1a antibodies and polyclonal IVIG on the activation status of vascular endothelial cells. Clin. Exp. Immunol. 137:216-222.
    View this article via: CrossRef PubMed Google Scholar
52. Armour, K.L., et al. 2006. Intravascular survival of red cells coated with a mutated human anti-D antibody engineered to lack destructive activity. Blood. 107:2619-2626.
    View this article via: CrossRef PubMed Google Scholar
53. Ghevaert, C., et al. 2007. HPA-1a antibody potency and bioactivity do not predict severity of fetomaternal alloimmune thrombocytopenia. Transfusion. 47:1296-1305.
    View this article via: CrossRef PubMed Google Scholar
54. Armstrong-Fisher, S., et al. 2004. In vitro materno-fetal transfer of native and Fc mutated recombinant antibodies. Vox Sang. 87(Suppl. 3):37.
55. Fanale, M.A., Younes, A. 2007. Monoclonal antibodies in the treatment of non-Hodgkin’s lymphoma. Drugs. 67:333-350.
    View this article via: CrossRef PubMed Google Scholar
56. Laemmli, U.K. 1970. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 227:680-685.
    View this article via: CrossRef PubMed Google Scholar
57. Bugert, P., et al. 2005. Microarray-based genotyping for blood groups: comparison of gene array and 5′-nuclease assay techniques with human platelet antigen as a model. Transfusion. 45:654-659.
    View this article via: CrossRef PubMed Google Scholar
58. Allen, D., et al. 2005. Collaborative study to establish the first international standard for quantitation of anti–HPA-1a. Vox Sang. 89:100-104.
    View this article via: CrossRef PubMed Google Scholar
59. Clark, M.R. 1997. IgG effector mechanisms. Chem. Immunol. 65:88-110.
    View this article via: PubMed Google Scholar