The Identification of the Amino Acids on Platelet Factor 4 That Bind Pathogenic Antibodies in Heparin-Induced Thrombocytopenia

Introduction: Heparin induced thrombocytopenia (HIT) is an adverse drug reaction that occurs when heparin binds to platelet factor 4 (PF4) and forms immunogenic multimolecular complexes. As a result, anti-PF4/heparin IgG antibodies bind to the PF4/heparin complexes, leading to cross-linking of FcγIIa receptors on platelets and FcγRI on monocytes resulting in their activation and an increased risk of thrombosis. Approximately 50% of cardiac surgery patients produce anti-PF4/heparin antibodies, but only a subset (<1%) produces pathogenic HIT antibodies. Current enzyme immunoassays (EIAs) cannot distinguish between pathogenic and non-pathogenic anti-PF4/heparin antibodies and will give a positive test even in the presence of the clinically insignificant non-pathogenic HIT antibodies. Further functional testing, such as the ¹⁴C-serotonin release assay (SRA), is required to identify samples containing the pathogenic anti-PF4/heparin antibodies that will lead to HIT. In this study, we hypothesized that different epitopes on PF4 could explain some of the differences between pathogenic and non-pathogenic antibodies.

Methods: We used alanine scanning mutagenesis (ASM) to produce 70 mutations of PF4, each with a unique amino acid substitution. Single point mutations were made by mutating non-alanine residues to alanine, and alanine residues to valine. The PF4 mutants were isolated from Escherichia coli and used in an EIA. The binding capacity of the monoclonal pathogenic HIT antibody (KKO) and a non-pathogenic anti-PF4/heparin antibody (RTO) to each of the PF4 mutants, relative to wild-type PF4, was determined. Point mutations of PF4 that resulted in >25% loss of binding were identified. We compared our candidate amino acids to binding sites previously identified in the PF4 and KKO crystal structure. In addition, we screened nine EIA-positive/SRA-positive (pathogenic) and six EIA-positive/SRA-negative (non-pathogenic) patient samples using the PF4 mutants. Student's t-test was used to determine whether the binding capacities of each amino acid was significantly different between the two patient sample groups.

Results: We identified 12 amino acid mutations of PF4 that caused ≥25% reduction in KKO binding. Most implicated amino acids cluster in two regions within the PF4 molecule. We identified amino acids on PF4 that were required for KKO binding which were not previously identified based on crystallographic studies. These results are consistent with the previous mutagenesis and crystallographic studies (Cai et al, Nat Commun. 2015;6:8277) of the KKO and PF4 complex but also contradict some previously stated amino acids that were thought to be part of this binding site. ASM of RTO and PF4 revealed only one amino acid that was affected. We used pathogenic and non-pathogenic EIA-positive human sera to further refine the location of the binding sites. We identified 10 amino acids that were required for the binding of human pathogenic HIT sera.

Conclusions: Using sequential point mutations, we identified the amino acids that are critical for the binding of pathogenic HIT antibodies to PF4. These amino acids are either directly involved in antibody binding or influence the antigenic conformation of PF4. This library of PF4 mutant proteins can help differentiate pathogenic and non-pathogenic HIT antibodies and this information can improve diagnostic assays.

This study was funded by the Canadian Institutes for Health Research.