Antibody Cloning Identifies Pathogenic and Non-Pathogenic Antibodies in Heparin-Induced Thrombocytopenia and Defines the Molecular Signatures That Differentiate the Two Types of Antibodies

Heparin-induced thrombocytopenia (HIT) is a common adverse drug reaction associated with frequent life-threatening thrombotic complications. The hallmark of HIT is polyclonal antibodies (Abs) that recognize platelet alpha granule chemokine PF4 when it binds to heparin (PF4/H). These Abs can be detected in solid phase assays that use PF4/H as a target (PF4 ELISA), but only a minority of patients testing positive actually have HIT, i.e., most heparin-induced Abs are non-pathogenic. In patients who have clinical HIT, Abs that activate platelets can be detected using a platelet-activation assay such as the serotonin release assay, or the PF4-dependent p-selectin expression assay (PEA) (Chest 2016; 150:506). Thus, there are at least two distinct types of heparin-induced Abs - those that react only in PF4 ELISA and are seemingly "non-platelet-activating" and "non-pathogenic" and those that are "platelet-activating" and "pathogenic". To date, the molecular basis for the differing clinical and serologic behaviors of pathogenic and non-pathogenic Abs is uncertain.

To address this issue, we performed single cell cloning to clone B cell receptors from IgG1⁺ B cells from HIT patients. We deposited single B cells (CD19⁺IgG1⁺) from 6 patients with "classical" and 2 patients with "spontaneous" HIT into 96 well plates containing feeder cells (from G Kelsoe, Duke U) that support B cell proliferation and Ab secretion (Immunity 2018;48:174). Clones secreting IgG were first screened in PF4 ELISA and positive results were obtained with 55 clones from 6 patients. Further screening showed that 7 of these clones (from 4 patients) were also PEA-positive (platelet-activating). Clones positive only in PF4 ELISA, positive in both PF4 ELISA and PEA, or negative in PF4 ELISA were designated NP (non-pathogenic), PA (platelet-activating) and NB (non-binding), respectively.

H and L chain variable regions were defined in 7 PA, 42 NP and 34 NB clones. The following findings were made when sequences in the 3 clonal groups were compared:

1. PA clones preferentially used JH6 (p=0.002) and the VH3/JH6 combination (p=0.0003)

2. The PA and NP Abs all employed κ chains, whereas κ chain usage for NB clones was 61% (p<0.0001).

3. No preferred signatures were identified in κ chain complementarity determining regions (LCDR3) of PA clones that differentiate them from NP and NB Abs.

4. PA Abs had longer heavy chain CDR3s (HCDR3) than NP (p<0.001) or NB (p=0.0001) Abs

5. PA Abs contained more positively charged amino acid residues compared to NP (p=0.058) or NB (p=0.002) Abs.

6. PA Abs contained more tyrosine residues compared to NP (p=0.067) or NB (p<0.0001) Abs

7. Five of 7 PA clones contained an RX_1-2K/RX_1-2R/H (RKH) motif in HCDR3; the remaining 2 PA clones contained a string of at least 5 tyrosines (Y₅ motif) in HCDR3. The RKH and Y₅ motifs were not found in any of the 76 NP and NB clones. Substitution of alanine for positively charged residues of the RKH motif or of tyrosine residues in the Y₅ motif in PA clones reduced PF4/H binding and platelet activation, arguing for functional significance of both motifs.

Utilization of nearly identical H and L chains within 3 groups of clones and of shared H chains within 3 groups of clones (both PA and NP) was observed in multiple patients. Moreover, utilization of a shared H chain was observed within 3 NP clones from two unrelated patients. These findings indicate clonal amplification and convergence of the B cell (both PA and NP) response, likely in response to a common antigen.

High throughput sequencing of IgG H chains were performed on peripheral blood mononuclear cells (PBMC) from 7 HIT patients and 3 healthy donors. Eleven of 1585 H chain sequences (0.69%) from HIT patients contained the RKH and 18 (1.1%) contained the Y₅ motif. In 3 healthy donors, 4 of 1418 H chain sequences (0.28%) contained RKH and none (0%) contained Y₅. The findings reflect amplification of B cells with receptors containing RKH and Y₅ motifs in HIT patients (p=0.1 for excess RKH and p<0.0001 for Y₅ in HIT).

These observations provide the first characterization of Ig structural motifs that are favored for selection in the humoral immune response leading to HIT and suggest that the RKH and Y₅ CDR3 motifs in particular may contribute importantly to Ab pathogenicity. Findings made are expected to facilitate further work to define features specific to "pathogenic" HIT Abs and, possibly, to identify genetic variants that predispose individuals to experience HIT.