Evidence against a protein in plasma that is a product of a factor XI mRNA splice variant missing exons 6 and 7

To the editor:

We read with interest the paper in Blood from Asselta et al¹  describing a form of human factor XI (fXI) encoded by an mRNA splice variant lacking exons 6 and 7 (fXI-Δ6/7). We observe 3 bands when plasma fXI purified by IgG affinity chromatography is size fractionated by nonreducing gel electrophoresis (Figure 1A). The dominant 160 kDa form contains 2 identical subunits connected by a disulfide bond.^2-4  The fainter band (70-80 kDa) migrating with reduced fXI represents single subunits. The third band, marked X in Figure 1A, is likely the approximately 105-kDa species observed in plasma immunoprecipitates by Asselta et al.¹  FXI subunits contain 4 apple domains (A1-A4, exons 3-10) and a protease domain (exons 11-15).^2,3  A fXI-Δ6/7 subunit would lack parts of A2 and A3 (amino acids 145-234). However, modeling (Figure 1B) indicates a complete apple domain could form from the N- and C-terminal portions of A2 and A3, respectively, encoded by exons 5 and 8.⁵  Asselta et al showed that fibroblasts transfected with fXI-Δ6/7 cDNA synthesize a protein that may have this hybrid domain,¹  and postulated that fXI-Δ6/7 and the 105-kDa form in plasma are identical based on electrophoretic mobility. Here we show that the 105-kDa plasma form is not likely to be fXI-Δ6/7.

First, the 160- and 105-kDa species are both secreted by fibroblasts transfected with full-length factor XI cDNA (Fig 1A).^6,7  Second, monoclonal antibodies that bind to the A3 (O1A6)^8,9  or A2 (14E11)⁹  domains both recognize all 3 fXI forms on Western blots (Figure 1C). We transfected HEK293 fibroblasts with expression constructs for wild type fXI (WT-fXI) and fXI-Δ6/7. In an enzyme-linked immunosorbent assay (ELISA), cell lysates for WT-fXI and fXI-Δ6/7 both gave strong signals when polyclonal anti–human fXI IgG or a monoclonal IgG against the protease domain (1G5)^6,7  is the capture antibody (Figure 1D). In contrast, only WT-fXI is captured by O1A6 or 14E11. Immunoprecipitates of cell lysates were studied by Western blotting. Again, polyclonal IgG recognized both WT-fXI and fXI-Δ6/7 (Figure 1E left panel), whereas O1A6 and 14E11 recognized only WT-fXI (Figure 1E center and right panels). These data clearly show that the 105-kDa plasma protein is not identical to the protein from cells transfected with the fXI-Δ6/7 cDNA.

The nature of the 105-kDa plasma species is not clear. It probably is a dimer, as it is not observed in reduced fXI (Figure 1A). Because both 160- and 105-kDa forms are secreted together by cells in serum-free media, they probably vary due to differences in posttranslational modifications. The apple domains of fXI form a tightly packed platform with extensive interdomain interfaces.^5,10  Our initial evaluation indicates the hybrid domain in fXI-Δ6/7 would not form interfaces properly, possibly resulting in an arrangement where the apple domains are like beads on a string. The disruption of the platform would likely interfere with binding to high-molecular-weight kininogen,¹⁰  a protein that forms a complex with fXI in plasma. This, and the absence of fXI coagulant activity,¹  raise questions about the function of such a protein, if it was expressed.