Anti-platelet factor 4/polyanion antibodies mediate a new mechanism of autoimmunity

Abstract

Antibodies recognizing complexes of the chemokine platelet factor 4 (PF4/CXCL4) and polyanions (P) opsonize PF4-coated bacteria hereby mediating bacterial host defense. A subset of these antibodies may activate platelets after binding to PF4/heparin complexes, causing the prothrombotic adverse drug reaction heparin-induced thrombocytopenia (HIT). In autoimmune-HIT, anti-PF4/P-antibodies activate platelets in the absence of heparin. Here we show that antibodies with binding forces of approximately 60-100 pN activate platelets in the presence of polyanions, while a subset of antibodies from autoimmune-HIT patients with binding forces ≥100 pN binds to PF4 alone in the absence of polyanions. These antibodies with high binding forces cluster PF4-molecules forming antigenic complexes which allow binding of polyanion-dependent anti-PF4/P-antibodies. The resulting immunocomplexes induce massive platelet activation in the absence of heparin. Antibody-mediated changes in endogenous proteins that trigger binding of otherwise non-pathogenic (or cofactor-dependent) antibodies may also be relevant in other antibody-mediated autoimmune disorders.

Figure 1. Dose-dependent binding of anti-PF4/P-ABS to PF4/H complexes in EIA.

PF4/H complexes were coated to a microtiter plate and binding of ABS to PF4/H complexes was measured by EIA. Mean OD values±s.d. were averaged from 5 sera for each group and 2 sera for control IgG. (a) The curve of control IgG (black stars) provides the background reaction. The specific reactions were lowest for group-1 (dark cyan circles), higher for group-2 (blue squares) and highest for group-3 (red triangles) ABS. (b) Slopes (s) of the curves obtained in (a) for concentrations up to 20 μg ml⁻¹ (at which the background for control IgG was an OD up to 0.32).

Figure 2. Platelet aggregates induced by purified PF4/P-ABS in the HIPA test.

(a) Dependence of platelet aggregation on ABS concentration: group-1 ABS (n=5) did not activate platelets, neither in the absence (−), nor in the presence (+) of reviparin up to a concentration of 89.7 μg ml⁻¹ (dark cyan); group-2 ABS (blue; n=5) induced platelet activation (red part) at concentrations ≥44 μg ml⁻¹ but only in the presence of reviparin. Group-3 ABS (n=5) (red) activated platelets at much lower concentrations (≥5 μg ml⁻¹) either in the presence or absence of reviparin. (b–o) SEM images show detailed platelet aggregates in the presence of different ABS. Only small aggregates occurred in the presence of control IgG (b–d) or group-1 ABS (e–g), regardless, whether or not reviparin was added. Group-2 ABS induced large platelet aggregates in the presence of reviparin after a lag-time of 15 min (h), but not earlier (i,j); while group-3 ABS activated platelets within 5 min independently of reviparin (k–m). Enlargements show looser platelet aggregates in the presence of group-2 ABS (n), but tighter and denser aggregates for group-3 ABS (o). Scale bar, 10 μm (b–m); 1 μm (n,o).

Figure 3. Binding strength of antibodies to PF4/H complexes measured by SMFS.

Experimental setup: (a) via PEG linkers, a single antibody is attached covalently on an AFM-tip, while the PF4/H complex is immobilized on the substrate. (b) The antibody interacts with the PF4/H complex when the tip approaches the substrate. (c) When the tip moves away from the substrate, the rupture force of the antibody from the complex is recorded. (d) Typical force–distance curve showing the rupture force (F) between group-3 ABS and the PF4/H complexes. (e) The average rupture force and the corresponding s.e. were determined by a Gaussian fit (solid curves) shown for one representative experiment for human control IgG (black), KKO (grey), group-1 ABS (dark cyan), group-2 ABS (blue) and group-3 ABS (red). (f) Average rupture forces and s.d. obtained from different cantilevers for KKO (grey; n=18), group-1 ABS (dark cyan; n=43), group-2 ABS (blue; n=54) and group-3 ABS (red; n=51). Group-2 and group-3 showed higher rupture forces than group-1 ABS and KKO. Rupture force between KKO and group-1 ABS did not differ significantly (P=0.877), while group-2 differ significantly from group-1 (P<0.001) and group-3 differ significantly from group-2 ABS (P=0.006).

Figure 4. Differences in binding characteristics of single antibodies.

To each cantilever, a different antibody purified from the same serum was attached (five sera per group). Each dot shows the mean and s.e. of the rupture forces for each respective antibody. (a) KKO and (b) group-1 ABS bind to PF4/H complexes with a binding strength mostly lower than 60 pN (black dotted line), while (c) group-2 and (d) group-3 ABS consist of ABS with different binding forces. A subset of group-3 ABS binds to PF4/H complexes with rupture forces higher than 100 pN (red dotted line). (e) Reverse experimental design: a PF4/H complex was immobilized on the tip, while antibodies were immobilized on the substrate (inset). The force histogram for control IgG (black) was used to determine the background. KKO (grey) and group-1 ABS (dark cyan) showed only one force distribution, while two distributions were observed for group-2 (blue) and group-3 (red) ABS. The second force distribution of group-3 ABS shifted to a higher force regime as compared to group-2 ABS (arrows).

Figure 5. Group-3 ABS cluster PF4.

(a–d) Representative binding isotherms for the titration of different antibody groups into the cells containing PF4: raw titration data (upper panel), and integrated heats (lower panel). (a) At concentration of 62.5 nM, the thermogram did not change for KKO, or (b) group-2 ABS (n=2), but (d) clearly changed for group-3 ABS (n=2). (c) At higher concentration, slight changes occurred in the thermogram measuring group-2 ABS at 950 nM (n=2), but no changes occurred in the thermogram for KKO at concentrations up to 6.67 μM. The scale for group-3 ABS in (d) is 1,000 times larger compared to the scale in a–c. (e) EIA ODs were lowest for group-1 ABS, higher for group-2 ABS, and highest for group-3 ABS when they interact either with PF4 (light grey) or with PF4/H complexes (dark grey). Binding of all antibodies is strongly enhanced by heparin. The red dotted line indicates the usual OD cutoff of the PF4/H EIA. (f) The number of binding counts was largely different when group-1 and group-2 ABS interacted either with PF4 (light grey) or with PF4/H complexes (dark grey) but did not differ for KKO and group-3 ABS. (g) DLS results show that PF4 and ABS sizes are smaller than 10 nm (black); when forming complexes with PF4 (blue) group-1 and group-2 ABS did not induce large aggregates while group-3 ABS formed complexes even larger than aggregates formed by heparin (grey). (h) The yield of antibodies purified by a PF4 column was highest for group-3 ABS as compared to other antibody groups, and only ABS purified from group-3 sera by the PF4 column were positive in the HIPA. The effluent of the column from group-2 and group-3 sera contained antibodies which activated platelets in the presence of heparin (data not shown).

Figure 6. PF4/group-3 ABS complexes expose the binding epitope for heparin-dependent ABS.

(a) Two schematics of either PF4/group-3 ABS complexes (1) or PF4/heparin complexes (2) bound to the tips interacting with KKO, group-1, or group-2 ABS immobilized on the substrates. (b) The rupture forces did not differ between these two systems, indicating that the group-3 ABS could mimic the effect of heparin on PF4 conformation. (c) Force histograms of the interaction of group-2 ABS with PF4/H complexes (top panel); or PF4/group-3 ABS complexes (bottom panel): ∼70% of binding forces were ≤60 pN and ∼30% of binding forces ≥60 pN (n=5 for each ABS group). (d) The summary counts of interactions with binding forces >60 pN between either PF4, PF4/H, or PF4/group-3 ABS complexes with KKO, group-1, or group-2 ABS again underscores that group-3 ABS can form complexes with PF4 which allow binding of heparin-dependent ABS. (e) DLS results show that when PF4/group-3 complexes (red) were incubated with different antibody groups, no significant changes in the sizes occurred when control IgG (black) or group-1 ABS (dark cyan) were added, complexes increased in size in the presence of group-2 ABS (blue) (P=0.047, ANOVA tested).