Pregnancy-specific glycoproteins bind integrin αIIbβ3 and inhibit the platelet-fibrinogen interaction

Abstract

Pregnancy-specific glycoproteins (PSGs) are immunoglobulin superfamily members encoded by multigene families in rodents and primates. In human pregnancy, PSGs are secreted by the syncytiotrophoblast, a fetal tissue, and reach a concentration of up to 400 ug/ml in the maternal bloodstream at term. Human and mouse PSGs induce release of anti-inflammatory cytokines such as IL-10 and TGFβ1 from monocytes, macrophages, and other cell types, suggesting an immunoregulatory function. RGD tri-peptide motifs in the majority of human PSGs suggest that they may function like snake venom disintegrins, which bind integrins and inhibit interactions with ligands. We noted that human PSG1 has a KGD, rather than an RGD motif. The presence of a KGD in barbourin, a platelet integrin αIIbβ3 antagonist found in snake venom, suggested that PSG1 may be a selective αIIbβ3 ligand. Here we show that human PSG1 binds αIIbβ3 and inhibits the platelet - fibrinogen interaction. Unexpectedly, however, the KGD is not critical as multiple PSG1 domains independently bind and inhibit αIIbβ3 function. Human PSG9 and mouse Psg23 are also inhibitory suggesting conservation of this function across primate and rodent PSG families. Our results suggest that in species with haemochorial placentation, in which maternal blood is in direct contact with fetal trophoblast, the high expression level of PSGs reflects a requirement to antagonise abundant (3 mg/ml) fibrinogen in the maternal circulation, which may be necessary to prevent platelet aggregation and thrombosis in the prothrombotic maternal environment of pregnancy.

Figure 1. Human and mouse PSGs inhibit the platelet – fibrinogen interaction.

PSG-mediated inhibition of the platelet – fibrinogen interaction was measured by estimating binding of Oregon Green-conjugated fibrinogen (OgFg) to washed human platelets using FACS. Fibrinogen binding to TRAP-activated platelets is set at 100% and resting platelets at 0%. All assays were analysed over a four or five point dose range of PSG proteins and mutants, from ∼5–100 or 200 µg/ml, depending on protein molecular weight. For clarity, some results are reported as single dose molar concentration comparisons between proteins. Protein molecular weights were calculated from amino acid sequences with no adjustments for posttranslational modifications. a, Binding of OgFg to human platelets in the presence of human CEACAM1, human IgG, and increasing doses of recombinant wildtype human PSG1. 4 µM PSG1 is equivalent to 200 µg/ml protein. b, Binding of OgFg to human platelets in the presence of (left to right): wildtype PSG1 (KGD); PSG1 in which the KGD tri-peptide motif is replaced with RGE, or AAA; PSG1 with deletion of N-domain; PSG1 N-domain; PSG1 N-domain in which the KGD tri-peptide motif is replaced with AAA. All proteins were used at 2 µM concentration, equivalent to 100 µg/ml full-length PSG1 variants, 75 µg/ml for PSG1ΔN, and 38 µg/ml for PSG1N variants. c & d, Binding of OgFg to human platelets in the presence of increasing concentrations of recombinant human PSG9 and mouse Psg23, respectively. 2 µM PSG9 and 2 µM Psg23 is equivalent to 100 µg/ml and 110 µg/ml, respectively. e, Summary of domain structures and mutants of PSG proteins used (see Fig. S3 in File S1 for sequences). f, Representative Coomassie-stained gels of protein used. For a - d, data are means of between three and seven independent experiments (detailed in main text) ± S.E.M. *, P<0.05; **, P<0.01; ***, P<0.001, nonparametric ANOVA with Dunnett’s multiple comparison post test.

Figure 2. Multiple domains of human PSG1 bind the platelet integrin αIIbβ3.

a, Integrin αIIbβ3 (2µg purified protein; lanes 1–3) pulls down PSG1 in an in vitro binding assay (lane 1). Negative controls are Protein G agarose beads with (lane 2) and without (lane 4) αIIbβ3, and with rabbit IgG instead of PSG1 (lane 3). Similarly, αIIbβ3 from lysates of CHO cell line stably transfected with αIIbβ3 (lanes 7, 8), but not lysate of sham transfected CHO control cell line (lanes 5, 6) pulls down PSG1 in co-immunoprecipitation assays. Negative controls lack PSG1, but contain α-αIIbβ3 mAb bound to beads (lanes 5 & 7). Western blotted membranes were probed with α-αIIbβ3 mAb Sz22 (upper gel) and α-PSG1 mAb-5 (lower gel). b, Commercial purified integrin αIIbβ3 bound to Protein G agarose beads pulls down recombinant PSG1 (lane 1) and PSG1ΔN (lane 2). Negative controls lack PSG1 (lane 3) or αIIbβ3 (lane 4). Western blotted membranes were probed with α-αIIbβ3 mAb Sz22 (upper gel), and α-His-Tag pAb (lower gel) which detects tagged PSG1 and PSG1ΔN proteins. c, Representative image and pooled data of fluorescent PSG1 (PSG1–800) binding to CHO cell line stably transfected with αIIbβ3 compared to sham transfected CHO control cell line. Cell density was measured using SYTO60. Data are means of six independent experiments ± S.E.M. *, P<0.05, Paired Student’s t-test. d, Binding of the activation-dependent monoclonal antibody, PAC-1, to platelet αIIbβ3. Washed human platelets were preincubated with BSA or PSG1 at 200 µg/ml before the addition of PAC-1 antibody and the indicated platelet agonist: TRAP (4 µM), thromboxane mimetic U46619 (250 nM), ADP (10 µM) or epinephrine (25 µM). Data are means of four independent experiments ± S.E.M. *, P<0.05, Student’s t-test. e, Washed platelets adhere and spread extensively on fibrinogen-coated (20 µg/ml) glass slides but poorly on 1% BSA-coated slides. Pre-incubation of platelets with 200 µg/ml PSG1 significantly reduced platelet adhesion and spreading on fibrinogen. Permeabilized platelets were stained for polymerized F-actin with Alexa-488 fluorescein isothiocyanate phalloidin before visualisation using confocal microscopy. Representative images are shown. Scale bar is 20 µm. Graph shows quantification of platelet adhesion as described in Methods. Data are means of three independent experiments ± S.E.M. *, P<0.05, Student’s t-test.

Figure 3. PSG1 does not activate platelets.

a, Washed human platelets were treated at 37°C for 3 min with TRAP (4 µM) and/or PSG1 (200 µg/ml) for 2 min as indicated. Alternatively platelets remained untreated (resting). Platelet activation was assessed by analysis of the phosphotyrosine profile by western blotting with the antiphosphotyrosine mAb 4G10. Experiment was performed twice. b, c & d, In a similar series of experiments, three different markers of platelet degranulation were assessed. For ADP secretion assay (b), platelets were treated with 4 µM TRAP and/or PSG1 (100 µg/ml) for 3 min at 37 ^OC. For surface expression of CD62P (c) and CD63 (d) platelets were treated with 4 µM TRAP and/or PSG1 (100 µg/ml) for 10 min at RT as described in Methods. Alternatively platelets remained untreated (resting). Data represent the means of three independent experiments ± S.E.M.

Figure 4. PSG1 is anti-thrombotic under arterial flow.

a & b, Representative images of adhesion of platelets under arterial flow following addition of 200 µg/ml PSG1 to 3 ml of circulating whole human blood. 200 µg/ml rabbit IgG was used as a negative control. Abciximab, an αIIbβ3 antagonist, was used as a positive control. b, Summary data of eight replicated independent arterial flow experiments expressed as means ± S.E.M. *, P<0.05; **, P<0.005; ***, P<0.0005 vs IgG, Mann Whitney test.