Human platelet antigen (HPA)-1a peptides do not reliably suppress anti-HPA-1a responses using a humanized severe combined immunodeficiency (SCID) mouse model

Abstract

Fetal and neonatal alloimmune thrombocytopenia (FNAIT) occurs most frequently when human platelet antigen (HPA)-1a-positive fetal platelets are destroyed by maternal HPA-1a immunoglobulin (Ig)G antibodies. Pregnancies at risk are treated by administration of high-dose intravenous Ig (IVIG) to women, but this is expensive and often not well tolerated. Peptide immunotherapy may be effective for ameliorating some allergic and autoimmune diseases. The HPA-1a/1b polymorphism is Leu/Pro33 on β3 integrin (CD61), and the anti-HPA-1a response is restricted to HPA-1b1b and HLA-DRB3*0101-positive pregnant women with an HPA-1a-positive fetus. We investigated whether or not HPA-1a antigen-specific peptides that formed the T cell epitope could reduce IgG anti-HPA-1a responses, using a mouse model we had developed previously. Peripheral blood mononuclear cells (PBMC) in blood donations from HPA-1a-immunized women were injected intraperitoneally (i.p.) into severe combined immunodeficient (SCID) mice with peptides and HPA-1a-positive platelets. Human anti-HPA-1a in murine plasma was quantitated at intervals up to 15 weeks. HPA-1a-specific T cells in PBMC were identified by proliferation assays. Using PBMC of three donors who had little T cell reactivity to HPA-1a peptides in vitro, stimulation of anti-HPA-1a responses by these peptides occurred in vivo. However, with a second donation from one of these women which, uniquely, had high HPA-1a-specific T cell proliferation in vitro, marked suppression of the anti-HPA-1a response by HPA-1a peptides occurred in vivo. HPA-1a peptide immunotherapy in this model depended upon reactivation of HPA-1a T cell responses in the donor. For FNAIT, we suggest that administration of antigen-specific peptides to pregnant women might cause either enhancement or reduction of pathogenic antibodies.

Introduction

Fetal and neonatal alloimmune thrombocytopenia (FNAIT) is due most commonly to destruction of fetal human platelet antigen (HPA)-1a-positive platelets by maternal HPA-1a antibodies. This results from platelet incompatibility between the HPA-1a-positive fetus, inheriting paternal antigens, and the HPA-1b1b mother. Untreated, the severity of clinical symptoms resulting from fetal platelet sequestration varies from mild purpura to intracranial haemorrhage, which in severe cases (when the fetal platelet count is <20 × 10⁹/ml) may result in fetal death or lifelong neurological sequelae [1]. FNAIT is the platelet equivalent of haemolytic disease of the fetus and newborn, but the major difference is the frequent occurrence of FNAIT in first pregnancies [2].

FNAIT occurs in approximately one in 1000–2000 live births [3,4], but is a significant obstetric problem that may require frequent antenatal treatments [5]. Intrauterine platelet transfusions (IUT) are effective, but can be associated with 1–2% fetal loss per procedure or 6% per pregnancy [6] and are now avoided, if possible. Administration of intravenous immunoglobulin (IVIG) and/or steroids to the mother are generally successful and are now the treatment of choice [7,8]. This is expensive, and often not well tolerated [9]. There are no reliable predictors of fetal disease and, as it is usually only recognized after the birth of an affected infant, antenatal management of subsequent pregnancies is guided mainly by measurement of maternal HPA-1a antibody and the outcome of previously affected pregnancies [7,10,11]. No current treatment can prevent immunization.

The HPA-1a/1b polymorphism results from a single amino acid substitution – Leu33 (HPA-1a) to Pro33 (HPA-1b) of the β-chain of the platelet integrin GPIIbIIIa (CD41/61) [12]. A rare third alloantigen, HPA-1 with Val33 [13], was recently designated HPA-1c [14]. The anti-HPA-1a immune response is highly restricted (>90%) to HPA-1b1b, human leucocyte antigen (HLA)-DRB3*0101 (HLA-DRw52a) women [15], indicating the presentation of HPA-1a peptides to T helper cells by this major histocompatibility complex (MHC) class II molecule. There is overwhelming evidence to suggest a major role for T cells in recognizing the peptide sequence containing Leu33 and the subsequent immune response leading to anti-HPA-1a formation and FNAIT [16–21].

There is evidence that specific peptides from antigens responsible for pathogenic immune responses can ameliorate or delay the onset of some autoimmune diseases and allergies [22–24]. Clinical development of peptide immunotherapy has been reported for multiple sclerosis [25], systemic lupus erythematosus [26], rheumatoid arthritis [27], diabetes [28] and bee [29] and cat allergy [30–32]. Murine models have demonstrated reduction of anti-tetanus toxoid (TT) [33], influenza haemagglutinin (HA) [34] and anti-D [35] by antigen-specific peptides. Intranasal (i.n.) administration of peptides induced tolerance in murine models of asthma [36] multiple sclerosis [37] and the K blood group antigen [38].

Because of the high restriction of anti-HPA-1a development to women with HLA-DRB3*0101, we investigated whether or not this specific antibody response could be susceptible to peptide immunotherapy. If successful, this might reduce the risks and costs associated with present treatments. We used a preclinical model [39] comprising severe combined immunodeficient (SCID) mice injected with human peripheral blood mononuclear cells (PBMC) from HPA-1a alloimmunized women (Hu-PBL-SCID) and HPA-1a-specific peptides followed by allogeneic HPA-1a-positive platelets. SCID mice do not reject xenogeneic cells, and tissues and human antibody responses can be followed for months [40,41] longer than other humanized murine models [42]. We had developed this method using normal donor PBMC and studied the effects of TT peptides on TT antibody responses [33,41] before applying it to rare FNAIT donors.

The HPA-1a peptides used in the present investigation varied from 12 to 22 mer and incorporated the three known anchor residues (Trp25, Asp28, Leu33) required for high-affinity binding to pockets 1, 4 and 9, respectively, in the peptide binding groove of the HLA-DRB3*0101 molecule, thus forming the T cell epitope [16,19–21,43].

All the elements involved in the anti-HPA-1a response in the Hu-PBL-SCID mouse model were human, including PBMC, antigenic platelets, natural HLA restricting molecules and antigen-specific T and B cells. Thus the immune responses observed should represent those that would occur in patients receiving peptide immunotherapy. In the majority of experiments it was found that HPA-1a peptides augmented rather than reduced anti-HPA-1a antibody responses.

Materials and methods

Donors

Seven donors were recruited for this study who had had babies diagnosed with FNAIT that resulted either in death of the fetus or mild neonatal bleeding (petechiae or purpura) due to low platelet counts (Table 1). These women were recruited from a previous study on detection of circulating HPA-1a-specific T cells during pregnancy [18] or from fetal medicine units in St Michael's Hospital (Bristol) or the John Radcliffe Hospital (Oxford), where they had received treatment as patients (IVIG and/or IUT of compatible platelets) during second and subsequent pregnancies. Recruitment and blood donation was made within the terms of the Ethical Approval of the project (04/Q2001/137) and policies and procedures of the National Health Service Blood and Transplant (NHSBT).

Table 1.

Clinical information on fetal and neonatal alloimmune thrombocytopenia (FNAIT) donors.

Donor and donation number	Time since last pregnancy (months) to donation	Gravidae/stillbirths	Antenatal treatment	Platelet count (×109/l) of last affected neonate	Anti-HPA-1a concentration in donor plasma (IU/ml)
1i	32	4/2	IVIG/IUT	6	426
1ii	37	4/2	IVIG/IUT	6	440
1iii	46	4/2	IVIG/IUT	6	417
2i	43	3/1	IVIG/IUT	20	37
2ii	52	3/1	IVIG/IUT	20	38
2iii	71	3/1	IVIG/IUT	20	38
3i	26	3/1	IVIG/IUT	50	9
3ii	34	3/1	IVIG/IUT	50	10
4i	59	3/1	IVIG/IUT	95	1·2
4ii	80	3/1	IVIG/IUT	95	1·7
5	49	2/0	IVIG/IUT	8	1·9
6	20	2/0	IVIG	7	13
7i	17	5/0	IVIG	2	35
7ii	30	5/0	IVIG	2	36

HPA: human platelet antigen; IVIG: intravenous immunoglobulin; IUT: intrauterine platelet transfusion.

Preparation of PBMC

Whole blood (∼450 ml) was collected into blood bags containing citrate, phosphate and dextrose (CPD). PBMC were isolated by density gradient centrifugation over Histopaque-1077 (Sigma-Aldrich, Poole, UK) and then washed in RPMI-1640 medium (Sigma-Aldrich) containing 10 IU/ml sodium heparin (CP Pharmaceuticals, Wrexham, UK), 100 IU/ml penicillin (P) (Sigma-Aldrich) and 100 μg/ml streptomycin (S) (Sigma-Aldrich). Two subsequent washes were without heparin. Recovered plasma was stored at −20°C. A sample of cells was removed for the T cell proliferation assay and another sample frozen (−80°C) for genotypic analysis. The remaining cells were centrifuged and resuspended in sterile phosphate-buffered saline (PBS) to 70–90 × 10⁶ cells/ml for injection into SCID mice.

Culture of human cells prior to injection into SCID mice with peptides

Two experiments were performed where PBMC were cultured for 48 h prior to injection into SCID mice. Cells were plated at 16 × 10⁶ cells/ml in RPMI-1640 with P, S and L-glutamine (200 mM) (Sigma-Aldrich) (cRPMI) and 5% (v/v) autologous heat-inactivated clarified human plasma (HIC-plasma). Recombinant HPA-1a antigen (IBGRL Reagents, Bristol, UK) at 10 μg/ml was added together with either 25 ng/ml interleukin (IL)-2 and 50 ng/ml IL-21 (Peprotech, London, UK) for experiment 2ii; or 1% (v/v) phytohaemagglutinin (PHA) (Murex, Dartford, UK) and 40 ng/ml IL-2 for experiment 4ii.

Culture of PBMC for cytokine responses in vivo and in vitro

PBMC were either cultured in cRPMI with or without Mycobacterium tuberculosis purified protein derivative (PPD) (50 μg/ml; Central Veterinary Laboratory, Addlestone, Surrey, UK) for 2 h, washed and then injected into SCID mice or were cultured in cRPMI and 5% (v/v) AB serum (ABS) (NBS Reagents, Liverpool, UK) (cRPMI/ABS) with or without PPD (50 μg/ml) for 17 days. Tail vein bleeds (TVB) were taken from the mice on days 2, 7 and 17 and supernatants were removed from cell cultures on the same days. Plasma from TVB from non-injected SCID mice and cRPMI/ABS were used as negative controls for the in vivo and in vitro tests, respectively. Concentrations of human IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, granulocyte–macrophage colony-stimulating factor (GM-CSF), interferon (IFN)-γ and tumour necrosis factor (TNF)-α were determined using a human 10-plex bead immunoassay kit (Invitrogen, Paisley, UK) and a LiquiChip 200 Luminex Dual Laser Detection System (Qiagen, Crawley, UK), following the manufacturers' instructions.

Genotypic analysis

PBMC (∼5 × 10⁶) from each donor were tested for the HLA-DRB3*0101 and HLA-DQB1*0201 alleles by polymerase chain reaction sequence-based typing (PCR-SBT) and HPA genotyping by PCR sequence-specific primers (SSP). This was performed by the Histocompatibility and Immunogenetics Laboratory (H&I Lab), NHSBT, Bristol, UK.

Platelets

HPA-1a-positive platelet donors were selected on the basis of their known HPA genotype from the NHSBT platelet donor database. Platelets from plateletpheresis donations were washed twice in sterile ethylenediamine tetraacetic acid (EDTA) buffer (20 mM NaCl, 30 mM Na₂HPO₄.2H₂O, 9 mM EDTA) with centrifugation at 3000 g for 10 min, resuspended at 1 × 10⁹ platelets/ml in platelet-freezing buffer [cRPMI (10% v/v), fetal calf serum (FCS) (80% v/v), dimethylsulphoxide (DMSO) (10% v/v) and 9 mM EDTA] and then stored in 1 ml aliquots at −80°C. When required, tubes were thawed quickly at 37°C, platelets washed twice in EDTA buffer and then resuspended in appropriate media.

Peptides

HPA-1 peptides, 12–22 mer, were synthesized at Bristol University using F-moc chemistry on resin, with 85–95% purity tested by high-performance liquid chromatography (HPLC) and amino acid analysis, lyophilized and stored in 1-ml aliquots at −20°C to minimize oxidation. The peptides were chosen from identification of the peptide sequences binding HLA-DRB3*0101 and their activation of specific T cells [16–21]. The amino acid sequences of the HPA-1a (Leu33) peptides were:

• HPA-1a₂₂ (24–45) 22 mer AWCSDEALPLGSPRCDLKENLI

• HPA-1a₂₀ (20–39) 20 mer SPMCAWCSDEALPLGSPRCD

• HPA-1a₁₆ (18–33) 16 mer AVSPMCAWCSDEALPL

• HPA-1a₁₄ (21–34) 14 mer PMCAWCSDEALPLG

• HPA-1a₁₂ (23–34) 12 mer CAWCSDEALPLG

The amino acids shown in bold type represent the known anchor residues for the HPA-1a T cell epitope: Trp25 (W), Asp28 (D) and Leu33 (L). The HPA-1b peptides were the same but with Leu33 (L) replaced by Pro33 (P). For one experiment, HPA-1c peptides with Val33 (V) were used.

Individual peptides were used in the T cell proliferation assay but equal mixtures (by weight) of the three longer peptides (22, 20, 16 mer), three shorter peptides (16, 14, 12 mer) or all five peptides were made for the in vivo experiments.

T cell proliferation assay (TCPA)

The assays were carried out in parallel with the in vivo experiments. They were based on an established method [44] and were modified for detection of anti-HPA-1a-specific T cells [18]. PBMC were incubated in cRPMI with 5% (v/v) HIC-plasma for 7 days and incorporation of [³H]-thymidine (Amersham International, Amersham, UK) was determined on days 4–7. Positive control antigens were 50 μg/ml PPD and 1 IU/ml TT (Evans Medical Ltd, Leatherhead, UK). The peptides were added to give a final concentration of 1, 3, 10 or 30 μg/ml in single-well cultures. Control wells (single wells) received an equal volume of cRPMI with 5% (v/v) HIC-plasma. Responses [counts per minute (cpm)] of test wells were expressed as the stimulation index (SI), defined as cpm (test)/cpm (control). Values of 3 SI and over were considered a positive response. The cumulative SI (cSI) was the summation of all the positive SI values for the 4 days testing for each antigen (PPD, TT or peptides).

SCID mice

CB17/lcr-Prkdc (CB-17 SCID) mice were bred and maintained in pathogen-free flexible film isolators. Experimental mice were transferred to air-filtered Scantainers and all procedures carried out in a laminar airflow walk-in cabinet. Blood samples were taken by TVB into heparinized capillary tubes (Richardson, Leicester, UK), transferred into microtubes and plasma removed after centrifugation. Plasma from the first blood sample was tested for murine IgG. Mice between 6–12 weeks of age and with murine IgG concentrations <10 μg/ml were used for experiments. Murine natural killer (NK) cells and macrophages, which reduce graft acceptance [45], were depleted by injecting 12 μl anti-asialo GM-1 (Wako Chemicals, Wako, USA) intraperitoneally (i.p.) on days −1, 0 and 7. The antibody was diluted in 100 μl sterile PBS for the days −1 and 7 doses and incorporated into the PBMC on day 0.

Injection of PBMC, platelets and peptides into SCID mice and blood sampling

The scheme of study is shown in Fig. 1. Groups of 5–10 mice were used. Human PBMC were injected i.p. in 500 μl of sterile PBS (35–45 × 10⁶ cells per mouse) with or without peptides (100 μg of each peptide was added to the PBS) on day 0. Subsequent peptide injections (100 μg of each peptide, mixed) were made i.p. in 300 μl sterile PBS on days 3, 6 and 9. Alternatively, mice received peptides intranasally (i.n.) (100 μg of each peptide mixed in 25 μl sterile PBS), using a standard anaesthesia method with isoflurane (Merial, Harlow, UK), developed by the Animal Service Unit at Bristol University. Mice were assigned into test and control groups in each experiment and received either HPA-1a,-1b or-1c peptides or sterile PBS. HPA-1a-positive platelets were injected i.p. in 200 μl sterile PBS (1 × 10⁹ platelets per mouse) on days 14 and 21 to all the mice to boost anti-HPA-1a responses. TVBs were taken as above at approximately 2-week intervals to monitor human IgG and anti-HPA-1a responses (Fig. 1). Fourteen experiments were performed using different combinations of peptides, routes of peptide administration or pretreatment of PBMC.

Figure 1.

Scheme of study in severe combined immunodeficient (SCID) mice. Tail vein bleeds (TVB) were made on day −2, the plasma isolated and used to quantitate murine immunoglobulin (Ig)G. Anti-asialo GM-1 antibody was injected intraperitoneally (i.p.) on days −1, 0 and 7 (hatched arrows) to deplete murine macrophages and natural killer (NK) cells. Human peripheral blood mononuclear cells (PBMC) were injected i.p. on day 0 (long arrow). Peptides were injected i.p. or inhaled intranasally (i.n.) on days 0, 3, 6 and 9 (short arrows). Human platelet antigen (HPA-1a⁺) platelets were injected i.p. on days 14 and 21 (thick arrows). Up to six TVB were taken at approximately 14-day intervals between days 14 and 105, the plasma isolated and used to quantitate human IgG and anti-HPA-1a.

Quantitation of IgG concentrations

Murine and human IgG concentrations were determined using affinity-isolated goat anti-murine or anti-human IgG (Sigma-Aldrich) as capture and peroxidase-conjugated reagents, with purified murine or human IgG (Sigma-Aldrich) as standards [46,47].

Modified monoclonal antibody immobilization of platelet antigen (mMAIPA) assay for the detection and quantitation of anti-HPA-1a

A modified version of the original MAIPA assay [48,49] was used to accommodate the small volumes of plasma (10–50 μl) retrieved by TVB from the mice and to exclude EDTA in buffers that reduces assay sensitivity [50]. Briefly, HPA-1a-positive platelets were incubated with plasma, washed, solubilized and the resulting clarified lysates added to enzyme-linked immunosorbent assay (ELISA) plates precoated with anti-GPIIbIIIa (CD41/61) (PAB-1 at 10 μg/ml; IBGRL Reagents). After incubation and washing, bound IgG was detected with anti-human IgG peroxidase. Anti-HPA-1a in plasma was quantitated by comparison of absorbance read on a Multiscan plate reader (Labsystems, Basingstoke, UK) with that of serial dilutions of an anti-HPA-1a standard (03/152; National Institute for Biological Standards and Control, Potters Bar, UK) and computed using Genesis (version 3) software (Labsystems).

Statistical analysis

Significant differences in total human IgG and specific IgG anti-HPA-1a antibody concentrations between control and treated mice were analysed using the Mann–Whitney U-test using GraphPad Instat version 3.0 for Windows 95 (GraphPad Software Inc., La Jolla, CA, USA).

Results

Genotyping results for HPA-1, HLA-DRB3*0101 and DQB1*0201

All seven FNAIT donors were HPA-1b1b and positive for both HLA alleles, concurring with a previous report of this HLA association [51].

Concentration of anti-HPA-1a in FNAIT donor plasma

Anti-HPA-1a concentrations were found to be very similar in donors bled on more than one occasion, indicating continued stimulation of the HPA-1a immune response. However, there was great variation in anti-HPA-1a concentrations between the donors, ranging from 1·2 to 440 IU/ml (Table 1).

T cell proliferation assay (TCPA)

The TCPA was performed to give an indication of the presence of HPA-1a and HPA-1b specific T cells in PBMC of every donation from the FNAIT donors.

With the control antigens, the greatest proliferation to PPD usually occurred on day 4, with that to TT being a little later on days 5–6, but by day 7 the SI with TT was always greater than the SI for PPD. This reflected the earlier secondary T cell responses to PPD than to TT observed in a previous study [18]. Maximum proliferation to HPA-1a (Leu33) peptides usually occurred on days 6 or 7 and may be indicative of delayed secondary responses. This was most obvious for two donations with the greatest responses (1ii and 7i). The concentrations of peptides did not, however, appear to show a consistent relationship with proliferation (Table 2).

Table 2.

T cell proliferation data for the donors with reactivity to human platelet antigen (HPA)-1a peptides.

Three donors (1, 2 and 4) had T cell responses to HPA-1a peptides on some occasions (Table 2) but not with other donations (Table 3), while donor 7 demonstrated responses to the peptides from both donations. Activity towards PPD and TT also varied; for example, PPD elicited greater proliferation than TT with 1ii, but this was reversed with 1i.

Table 3.

Summary of T cell proliferation assay (TCPA) data and human peripheral blood lymphocytes-severe combined immunodeficient (Hu-PBL-SCID mouse model) experimental information, results and overall effect of peptides.

The three longer HPA-1a peptides (22, 20, 16 mer) elicited the most proliferation (Table 2). PBMC of two donors responded briefly to HPA-1a₁₄, but no proliferation was induced by the shortest 12 mer peptide. Most of the proliferative responses to HPA-1 peptides (11 of 15) were from donors 1 and 7; the 22 and 16 mer peptides were stimulatory for donor 1 PBMC, whereas the majority of responses from donor 7 PBMC were with the 20 mer peptide. Donors 1 and 7 also had high proliferation to HPA-1b₂₂ peptides. No other reactivities to the five HPA-1b peptides were obtained from these donors. When the five HPA-1c peptides were tested with donation 7ii, the 20 mer peptide elicited proliferation in two wells.

Antibody responses in Hu-PBL-SCID mice

After injection of PBMC 86% of the mice produced >0·25 mg/ml human IgG, which was considered successful reconstitution, indicating survival and proliferation of human cells. In these reconstituted mice, mean human IgG concentrations in murine plasma reached >1 mg/ml in 13 of 14 experiments, approximately 10% of the concentration in normal human serum, but were >5 mg/ml in four experiments (Figs 2e,g and 3a,e).

Figure 2.

Human antibody responses in human peripheral blood lymphocytes-severe combined immunodeficient (Hu-PBL-SCID) mice. Total immunoglobulin (Ig)G (a,c,e,g) and IgG anti-human platelet antigen (HPA)-1a concentrations (b,d,f,h) in plasma of mice reconstituted with peripheral blood mononuclear cells (PBMC) from four fetal and neonatal alloimmune thrombocytopenia (FNAIT) donations. Mice received HPA-1a (or-1c) peptides (▪), HPA-1b peptides (○) or phosphate-buffered saline (PBS) (x). (a,b) Donation 1i, 22 + 20 + 16 mer peptides intraperitoneally (i.p.), six mice per group. (c,d) Donation 1iii, 22 + 20 + 16 + 14 + 12 mer peptides intranasally (i.n.), six mice per group. (e,f) Donation 2i, 16 + 14 + 12 mer peptides i.n., four mice per group. (g,h) Donation 7ii, 22 + 20 + 16 + 14 + 12 mer HPA-1c peptides i.p., seven mice per group (▪). The mean ± standard error of the mean (s.e.m.) is shown. *Significant difference in values between the two groups, where P < 0·05.

Figure 3.

Human antibody responses in human peripheral blood lymphocytes-severe combined immunodeficient (Hu-PBL-SCID) mice. Total immunoglobulin (Ig)G (a,c,e,g) and IgG anti-human platelet antigen (HPA)-1a concentrations (b,d,f,h) in plasma of mice reconstituted with PBMC from four fetal and neonatal alloimmune thrombocytopenia (FNAIT) donations. Mice received 22 + 20 + 16 + 14 + 12 mer HPA-1a peptides intraperitoneally (i.p.) (▪), or 22 + 20 + 16 + 14 + 12 mer HPA-1b peptides i.p. (○). (a,b) Donation 2iii, four mice per group. (c,d) Donation 7i, seven mice per group. (e,f) donation 6, six mice per group. (g,h) donation 1ii, five mice per group. The mean ± standard error of the mean (s.e.m.) is shown.

PBMC of four donors (1, 2, 6 and 7) who had >10 IU/ml anti-HPA-1a in their plasma initiated anti-HPA-1a responses after injection into SCID mice (Figs 2 and 3). The presence of HPA-1a specific T cells in donor PBMC was also related to the development of specific antibody responses. The greatest anti-HPA-1a concentrations in control mice (>1 IU/ml) (Figs 2b,h and 3d,h) were reached using PBMC from donors 1 and 7, which had high proliferation to HPA-1a peptides with a cSI of ≥20. When there was little or no proliferation, no or low anti-HPA-1a (<1 IU/ml) was detected in mice (Figs 2d,f and 3b,f). No anti-HPA-1a was detected in mice of six experiments (data not shown) when the donors had either <10 IU/ml anti-HPA-1a in their plasma or no proliferative responses to HPA-1 peptides.

In eight experiments when anti-HPA-1a was detected in murine plasma all the reconstituted mice (i.e. with >0·25 mg/ml IgG) in the experimental groups were compared for the effect of peptides on antibody concentration. Some of these mice (18%) did not produce anti-HPA-1a but were included in the statistical analysis, which resulted in anti-HPA-1a responses between treated and control groups being not significantly different at P < 0·05 in all the experiments. A significant difference was observed in only one experiment between the IgG concentration in mice injected with HPA-1a peptides compared to PBS (Fig. 2a).

Anti-HPA-1a was present in murine plasma by 28 days in five of eight experiments. These specific antibody responses developed later than total IgG and usually waned earlier (Figs 2 and 3), except for one early, short and weak response (Fig. 2d). Anti-HPA-1a responses were followed for 36–93 days (mean 63 days) in the eight informative experiments.

The effects of HPA-1a peptides on anti-HPA-1a responses in Hu-PBL-SCID mice

The longer peptide mix (22, 20, 16 mer) was used initially but found to be stimulatory for the first experiment, with increased anti-HPA-1a concentrations at 42–56 days in mice receiving these peptides compared to the PBS control (Fig. 2b, donation 1i). Two further similar experiments were not informative because no anti-HPA-1a was produced. Peptide mixtures containing short 14 and 12 mer peptides which had stimulated little T cell proliferation were then tested with the following modifications to the protocol, but there was no evidence for their effectiveness in these four experiments.

When peptides were delivered via the i.n. route using PBMC from two donors with high plasma anti-HPA-1a (1iii with 417 IU/ml and 2i with 37 IU/ml) only low, similar concentrations of anti-HPA-1a were produced in the mice (Fig. 2d and f). These donations were found to have no HPA-1a-specific proliferative T cells (data not shown).

From a previous study it was found that culture of PBMC for 48 h prior to injection into SCID mice increased IgG and anti-TT antibody responses [41]. PBMC from two FNAIT donors (2ii and 4ii), who in previous experiments had produced no or low anti-HPA-1a, were therefore cultured in vitro with antigen and cytokines for 48 h before injection into mice. However, no anti-HPA-1a was detected in mice which correlated with the low proliferative responses to HPA-1a peptides in the TCPA (Table 3).

In eight experiments with six donors given a mix of all five peptides, differing results were obtained (Table 3). In one experiment there was no effect after i.n. administration of peptides (Fig. 2d, 1iii). Three other experiments were not informative (donations 2ii, 3ii, 5). With two donations (2iii and 7i) strong stimulation of anti-HPA-1a responses in vivo occurred (Fig. 3b and d). In contrast, an experiment with donor 6 was suggestive of peptide inhibition, with very low anti-HPA-1a responses only in control mice, near the limit of detection (Fig. 3f). For these seven donations, little or no T cell activity towards HPA-1 peptides was detected in the TCPA (Table 3). However, in one experiment (1ii), inhibition of the anti-HPA-1a response was mediated by the HPA-1a peptides, with reduction of anti-HPA-1a concentrations in all mice in the test group (Fig. 3h). Interestingly, high HPA-1a-specific T cell proliferation occurred for this donation.

Thus three donations from donor 1 gave differing results in vivo. Peptide inhibition of anti-HPA-1a occurred with high (364 cSI) HPA-1a-specific T cell proliferation (Table 2, 1ii). Peptide stimulation of anti-HPA-1a resulted when there were only few HPA-1a-specific T cells (20 cSI) (Table 2, 1i). Without HPA-1a reactive circulating T cells, little anti-HPA-1a was produced and the peptides had no effect (data not shown).

Furthermore, one experiment was performed with five HPA-1c peptides containing Val33 instead of Leu33 with donor 7. These peptides stimulated both T cell proliferative responses and B cell antibody responses. Anti-HPA-1a concentrations in SCID mice were 10 times greater with HPA-1c peptides compared to control HPA-1b peptides (Fig. 2h, Table 3).

Overall, of four informative experiments, inhibition of anti-HPA-1a responses by HPA-1a peptides occurred definitively in one, whereas in the other three experiments (one with the same donor) these peptides substantially enhanced anti-HPA-1a antibody responses.

Cytokine production by PBMC in SCID mice or in culture

The predominant human cytokines identified in plasma of six SCID mice 2 days after injection of PBMC were IFN-γ, IL-6 and IL-8. Addition of PPD did not affect these transient cytokine responses, but induced low concentrations of IL-5 at 7 days and IL-8 at 17 days. However, when the same donor PBMC were cultured in parallel, PPD stimulated production of GM-CSF, IL-1β and TNF-α and enhanced the synthesis of IL-6 and IL-8. Apart from TNF-α, these cytokine responses were sustained for the 17-day period in vitro (Fig. 4). IFN-γ was not produced in vitro. No IL-2, IL-4 or IL-10 was detected in any samples of murine plasma or culture supernatants.

Figure 4.

Analysis of cytokines produced by human peripheral blood mononuclear cells (PBMC) in (a) human peripheral blood lymphocytes-severe combined immunodeficient (Hu-PBL-SCID) mice and (b) in culture. Some PBMC were incubated with purified protein derivative (PPD) for 2 h prior to injection into mice, or cultured with PPD for 17 days. Tail vein bleeds (TVB) were taken from mice and samples of supernatants from PBMC on days 2, 7 and 17.

Discussion

Anti-HPA-1a was secreted de novo for 21–70 days by human PBMC in SCID mice with similar response kinetics to earlier reports of human antibody synthesis in this model [18,33,38,41]. PBMC produced a brief mildly inflammatory T helper type 1 (Th1) cytokine response detected 2 days after injection, but not at 7 or 17 days. This was unaffected by PPD, which induces Th1 cytokines and an inflammatory response. In contrast, cultured PBMC had an inflammatory cytokine profile that was enhanced by PPD. Thus, the human cytokines in vivo may have been favourable for anti-HPA-1a responses. However, these did not always occur.

Induction of an antibody response requires cognate interactions of antigen-specific T cells first with DC presenting antigen-specific peptides and then with antigen-specific B cells, also presenting these peptides on HLA. The frequency of HPA-1a-specific T and B cells would be low in the PBMC of FNAIT donors 1·5–7 years after immunization, in the region of <1 in 10 000 [52]. Thus, approximately 10³ of these cells would be in each mouse, dispersed first into the peritoneal cavity and then by 5 weeks to diverse tissues of the host [41]. If any of the steps in the immune response pathway failed, no anti-HPA-1a would be made. This antibody was not detected in 20% of the successfully reconstituted mice in the eight experiments described where the majority of mice harboured anti-HPA-1a responses (Figs 2 and 3). It was not possible to distinguish whether or not lack of anti-HPA-1a was due to an ineffective immune response or the effect of peptides. However, inclusion into the statistical analysis of the mice with zero anti-HPA-1a but with successful reconstitution (>0·25 mg/ml IgG) resulted in no significant differences between test and control groups. Nevertheless, the effect of peptides was apparent.

HPA-1a peptides stimulated anti-HPA-1a responses in three experiments, yet reduced them in one. With PBMC of one donor, either enhancement or amelioration of specific antibody production was obtained on two occasions. The only difference found between these donations was in the magnitude of HPA-1a-specific T cell responses in PBMC observed in the TCPA.

Augmentation of specific antibody responses by HPA-1a peptides occurred when few HPA-1a-specific T cells were detected in donors' PBMC, suggesting activation of naive or low-avidity T cells by the peptides. In contrast, in the single experiment when HPA-1a peptides mediated reduction of anti-HPA-1a (Fig. 3h), high HPA-1a-specific T cell activity in vitro was observed (Tables 2 and 3). Thus HPA-1a peptide inhibition was not donor-dependent but appeared to be related to the presence of high numbers of circulating HPA-1a-specific T cells.

We tested the possibility that HPA-1a peptides could have ligated HPA-1a-specific B cells directly, through the BCR, by utilizing altered peptide ligands with Val33 (HPA-1c) in place of Leu33 (HPA-1a). This rare HPA-1 allele was discovered during sequencing a large series of patients. Integrin αIIbβ3 (CD41/61) transfectants with β3-Val33 did not bind anti-HPA-1a from most (78%) FNAIT patients, whereas the wild-type Leu33 variant bound all these sera [13], thus Val33 did not usually provide the B cell epitope. Therefore, because we found HPA-1 peptides with Val33 were as effective as those with Leu33 at stimulating anti-HPA-1a production (Fig. 2h), it is likely that they acted on T cells to drive the antibody response. This conclusion accords with earlier findings that HPA-1 peptides with Val33 were as effective as HPA-1a peptides at stimulating proliferation of HPA-1a-specific T cell clones [21] and we found that both Val33 and Leu33 induced proliferation of cells in PBMC with donor 7 (Table 2).

The presence of HPA-1a-specific T cells in the peripheral circulation fluctuated unpredictably over time with donors 1, 2 and 4 in this study (Table 2), confirming earlier observations when three of six donations from another HPA-1a immunized woman had HPA-1a-specific T cells [18]. Such dynamic changes in T cell responses were also found to myelin proteolipid protein (PLP) [53], myelin basic protein (MBP) [54] and MBP, PLP and myelin oligodendrocyte glycoprotein (MOG) peptides [55] of patients with multiple sclerosis. This may reflect periodic reactivation of antigen-specific T cells. T cell responses to PPD and TT also varied, independently, in different donations from individuals ([18,33] and current study).

In contrast to T cells, anti-HPA-1a in donor plasma was present at remarkably similar concentrations over 14–28-month periods, up to 7 years after the last antigen stimulus (pregnancy). Long-term production of anti-HPA-1a has been reported previously [56,57] and is unlike other antibodies such as anti-D [58,59] or anti-TT [60], which reduce in concentration during these time-periods.

Other features of the anti-HPA-1a response are unusual. First, FNAIT due to anti-HPA-1a often occurs in primigravidae (>50% of cases) [2], whereas haemolytic disease of the newborn (HDFN) seldom does (6% of cases) [61]. Secondly, anti-HPA-1a is the most common platelet antibody causing FNAIT [2,62], but is rarely formed after transfusion [63]. Therefore, pregnancy-related antigenic material, possibly placental syncytiotrophoblast expressing αVβ3 (CD51/61) integrin, may cause the initial HPA-1a immunization [64–66]. Thirdly, HPA-1a alloimmunized women occasionally have weak T cell responses to HPA-1b peptides, found previously for 62% of pregnant FNAIT patients [18], 36% of non-pregnant women [17], the >57-year-old female donor [18] and in the current study for two donors. Fourthly, HLA restriction occurs with the anti-HPA-1a response but not anti-D [67]. HLA DRB3*0101 is associated with many autoimmune diseases [19]. Finally, IVIG is a successful therapy for many autoimmune diseases and FNAIT but is of low efficacy for ameliorating HDFN [68]. These features suggest that the anti-HPA-1a immune response and FNAIT have many characteristics of autoimmunity.

Peptide immunotherapy for autoimmune diseases has attracted much interest and research, although translation of findings from animal models into clinical effectiveness has been slow [24]. In contrast, treatment of allergic patients with allergen-specific peptides is proving remarkably successful [69]. The difference in outcome may be related to the nature of the immunogenic material, either self-antigens (possibly primed by unknown foreign antigens) causing autoimmunity or foreign proteins eliciting allergies. TT is a foreign protein and the anti-TT response, like those to allergens, is not HLA-restricted [70]. Using Hu-PBL-SCID mice, we found that immunodominant TT peptides caused amelioration of anti-TT responses [33]. This contrasts with the failure of HPA-1a peptides to consistently prevent anti-HPA-1a responses unless antigen-specific T cells were undergoing a cycle of reactivation in the donor. Therefore, peptide-specific immunotherapy might have different outcomes for modulating ongoing antibody responses to foreign proteins (TT) or the human HPA-1a glycoprotein alloantigen.

A strong ongoing anti-HPA-1a immune response in donors was necessary for anti-HPA-1a responses to develop in SCID mice. Despite these restrictions and the difficulties in generating data, the Hu-PBL-SCID mouse model is, to our knowledge, the only one yet devised in which all the human cellular and molecular components of the HPA-1a immune response are present, i.e. HLA DRB3*0101 on APCs, antigenic HPA-1a on human platelet GPIIbIIIa (CD41/61) and HPA-1a-specific T and B cells.

PBMC were well tolerated in SCID mice, and it was unlikely that xenogenicity affected the immune responses studied. Although graft-versus-host disease (GVHD) has been reported in irradiated Hu-PBMC-SCID mice [45] and irradiated IL-2Rγ-deficient Hu-PBMC-NOD-SCID mice [71], we did not observe this disease here or in earlier studies using Hu-PBMC-SCID mice [33,40,41,72].

Is it likely that peptide immunotherapy in pregnant women at risk of FNAIT would be more effective than our experiments suggested? More pregnant women (74%) had HPA-1a-specific T cells [18] than the non-pregnant women in the current study (50%), due possibly to repeated antigenic stimulation through fetomaternal haemorrhages [73], and therefore they may be more susceptible to HPA-1a peptide-induced tolerance. The immune system of pregnant women is skewed towards a mild proinflammatory Th2 profile [74–76]. Conversely, the Th phenotype of HPA-1a-specific T cell clones isolated from alloimmunized women [20,21] and of cultured PBMC from non-pregnant FNAIT donors [77] was Th1. This may make attempts at skewing of the immune response by peptides towards a regulatory phenotype too challenging. Also, pregnancy-associated mild systemic inflammation might cause activation of antigen-presenting cells (APCs) when they need to be quiescent for peptide-induced Th cell anergy [78,79]. In addition, the HPA-1a immune response differs among patients. T cell activation by HPA-1a peptides was variable [17,18,20,21] (Table 2). T cell clones had dissimilar TCR genes [20,21] that could affect their affinity for peptides. Anti-HPA-1a antibodies are known to be heterogeneous for their epitope specificity [80] and functional activity [81], and the large range in anti-HPA-1a concentrations we found (Table 1) may reflect differences in IgG affinity maturation. Importantly, unpredictable fluctuations in circulating HPA-1a-specific T cells occur in both pregnant [18] and non-pregnant FNAIT women (Table 3).

If, as our experimental results suggested, immunomodulation by HPA-1a peptides depended upon the presence of strong HPA-1a-specific T cell immunity, then it might not be possible to predict whether these peptides would inhibit or stimulate HPA-1a antibody responses in pregnant women. Our study, using a preclinical translational model, indicated that peptide immunotherapy is unlikely to reduce maternal alloantibodies reliably and prevent FNAIT.

Disclosure

The authors have no financial conflicts of interest.

Author contributions

D. J. performed the experiments, J. E. assisted with cytokine analysis, D. J. and B. K. designed the study and wrote the manuscript. The work was funded by NHS Blood and Transplant, UK.