FVIII Protein Is Not Detectable in Human PBMCs or Livers from Dogs with an Intron-22 Inversion Mutation: Implications for FVIII Immunogenicity and Tolerance

A clinically relevant question is whether partial FVIII proteins are expressed in tissues from patients with severe hemophilia A (HA) due to an intron-22 inversion mutation. If so, this could in principle confer central immune tolerance to the expressed FVIII regions, thereby lowering the risk of anti-FVIII immune responses. In 2013, Pandey et al. reported detection of FVIII proteins in peripheral blood mononuclear cells (PBMCs), B cells and liver tissues from an intron-22 inversion subject and non-HA controls. They concluded that partial FVIII proteins were translated from inverted mRNA encoding F8 exons 1-22, and also from the F8B transcript, which contains F8 exons 23-26 (Nature Medicine 19(10), 1318-24). In 2014, the Montgomery and Ginsburg labs reported that FVIII protein is expressed principally, and possibly exclusively, in endothelial cells (ECs). These studies utilized mouse models with EC-directed deletion of F8 exons 17-18, or of the FVIII transporter protein LMAN1, respectively. The F8-EC-knockout mice had a severe hemophilia A (HA) phenotype. Earlier studies from several labs, including Pandey et al., relying primarily on antibody staining, had reported FVIII expression in additional cell types, including PBMCs and hepatocytes.

The present study tests the hypothesis that partial FVIII proteins are expressed from inverted mRNA encoding F8 exons 1-22 and/or from F8B mRNA. A panel of FVIII-specific polyclonal and monoclonal antibodies (mAbs) was tested for FVIII specificity by staining permeabilized PBMCs, monocytes, T cells, blood outgrowth ECs, monocyte-derived macrophages and dendritic cells, HUVECs and B-cell lines. Positive and negative controls included FVIII-expressing BHK cells and non-engineered BHK cells. Specificity was further tested by immunoprecipitation (IP) of cell lysates followed by SDS-PAGE, Western blots and mass spectrometry to define proteins pulled down by the anti-FVIII antibodies. In addition, immunohistochemistry (IHC) staining was carried out for liver sections from HA dogs with an intron-22 inversion mutation and from otherwise normal (non-HA) control dogs.

IP followed by SDS-PAGE, Westerns and/or mass spec revealed that a significant fraction of the anti-FVIII antibodies bound to other proteins besides FVIII in the tissues and cells examined. A cocktail of 4 anti-FVIII mAbs that were validated using (+) and (-) control samples was used to stain isolated and cultured cells and tissues. Cross-recognition of canine FVIII was confirmed by IP + Western blots. IHC using the mAb cocktail followed by HRP-conjugated secondary antibodies produced variable staining of liver tissues from HA and non-HA dogs, regardless of tissue preparation methods. However, use of fluorescent-labeled secondary mAbs produced signals well above background in ECs of control dog livers but not in the HA livers. Multiple confocal images were selected randomly, and average MFI values of 10 non-HA liver sections were well above those of 10 HA sections (p = 0.006). A second staining method, the Duolink Proximity Ligation Assay, was employed as an independent test to detect interactions between FVIII and VWF. Again, only normal control liver sections showed fluorescence above background (p =0.0004). F8B mRNA was not detected in canine liver and was not expected based on the canine intron-22 DNA sequence. We conclude that if protein is expressed from inverted mRNA encoding F8 exons 1-22, it is below the detection limit of these assays.

Permeabilized human cells were tested for FVIII expression by staining using the mAb cocktail, or validated anti-FVIII-C2 mAbs to detect the putative protein encoded by F8B, followed by fluorescent detection. Cells were also stimulated with histamine followed by a chromogenic FVIII activity assay of supernatants. Results indicated that if FVIII or F8B protein are expressed in non-ECs, they are below the detection limit of these assays. We also note that if the putative F8B-encoded protein confers tolerance, immune responses of HA patients to the FVIII C2 domain would be quite rare, which is not the case.

Although neonatal thymic expression of partial FVIII proteins may occur, immune tolerance to self-antigens requires repeated exposures of the immune system to the antigens. We conclude it is unlikely that HA patients with an intron-22 inversion mutation have universally acquired central tolerance to partial proteins encoded by inverted F8 mRNA or F8B.