Identification of 31 novel mutations in the F8 gene in Spanish hemophilia A patients: structural analysis of 20 missense mutations suggests new intermolecular binding sites

Abstract

Hemophilia A (HA) is an X-linked bleeding disorder caused by a wide variety of mutations in the factor 8 (F8) gene, leading to absent or deficient factor VIII (FVIII). We analyzed the F8 gene of 267 unrelated Spanish patients with HA. After excluding patients with the common intron-1 and intron-22 inversions and large deletions, we detected 137 individuals with small mutations, 31 of which had not been reported previously. Eleven of these were nonsense, frameshift, and splicing mutations, whereas 20 were missense changes. We assessed the impact of the 20 substitutions based on currently available information about FV and FVIII structure and function relationship, including previously reported results of replacements at these and topologically equivalent positions. Although most changes are likely to cause gross structural perturbations and concomitant cofactor instability, p.Ala375Ser is predicted to affect cofactor activation. Finally, 3 further mutations (p.Pro64Arg, p.Gly494Val, and p.Asp2267Gly) appear to affect cofactor interactions with its carrier protein, von Willebrand factor, with the scavenger receptor low-density lipoprotein receptor-related protein (LRP), and/or with the substrate of the FVIIIapi*FIXa (Xase) complex, factor X. Characterization of these novel mutations is important for adequate genetic counseling in HA families, but also contributes to a better understanding of FVIII structure-function relationship.

Figure 1

Structure and function of Xase complex. (A) Schematic representation of Xase complex and mechanism of factor X activation. The major structural domains of cofactor VIIIa, its cognate protease FIXa, and substrate FX are labeled, and represented in the approximate positions they would occupy in the Xase complex. The long FX activation peptide is also indicated as a ribbon, pointing to the insertion of the Arg194(15)-Ile195(16) activation peptide bond in the active site of FIXa. (Numbers in parentheses refer to the standard chymotrypsinogen numbering system.) Gla indicates γ-carboxyglutamic–rich domain; EGF1/EGF2, epidermal growth factor–like domains 1 and 2; and SP, serine protease domain. (B) Three-dimensional model of human FVIIIa, highlighting the side chains of all novel missense mutations identified in the current investigation (light magenta spheres). The 5 FVIIIa domains are represented with their major secondary structure elements and color-coded as in panel A. For clarity, only selected residues are labeled, including the 4 exposed residues Arg64, Ala375, Gly494, and Asp2267 (boxed). The acidic a1 and a2 peptides are included only to indicate their locations on opposite poles of domain A2, as no appropriate templates are available for them. Selected side chains of hydrophobic residues in C1/C2 domains that would associate with the phospholipid membrane are in green. Some previously detected mutations of exposed residues are shown as yellow sticks; for clarity, only a few of these residues are labeled (see “Discussion” for details and references). Considering these previous analyses of FVIII mutants, residues important for cognate FIXa binding cluster to the right in the chosen orientation (eg, stretches Ser558-Gln565, Asp712/Lys713, and Glu1811-Lys1819), while those involved in FX recognition map to the left in this orientation (see, eg, the location of the acidic a1 peptide). This clear segregation of residues critical for FIXa/FX binding allows us to predict important roles for residues Pro64, Gly494, and Asp2267 in substrate binding and presentation to the FVIIIapi•FIXa (Xase) complex.

Figure 2

Multiple partial alignments of factor V, factor VIII, and ceruloplasmin from different species around FVIII positions where novel mutations were found in the current work. For simplicity, only mutations within domain A1 and the A1-A2 linker (part A) and domains C1/C2 (part B) are represented; a complete alignment is available from the authors upon request. Strictly conserved residues are white with black shading, and conservative changes are shaded gray. Numberings refer to the mature human proteins. The activation cleavage site is indicated with an arrow in panel A, and loops important for membrane association are boxed in panel B. The secondary structure elements given below FV and ceruloplasmin sequences correspond to the crystal structures of bovine inactivated FVa (PDB 1SDD, Adams et al), FVIII C2 domain (1D7P, Pratt et al), and human ceruloplasmin (1KCW, Zaitseva et al), as deposited in the corresponding PDB entries. Residues that are disordered in the crystal structure of ceruloplasmin are underlined.

Figure 3

Close-up of novel, putative type I missense mutations identified in the current study. Atoms are color-coded (green indicates carbon; blue, nitrogen; red, oxygen; and yellow, sulfur), and the mutated residues are labeled yellow. With exception of Gly261 (E) and His2082 (K), only residues within 4 Å of the mutation are shown. Hydrogen bonds are indicated with dotted lines. Notice that most affected residues are fully buried in the protein core and engage in multiple interactions with surrounding residues. (Further explanation for each mutation is in “Type I mutations.”)

Figure 4

Comparison of loop structures around FVIII C2 residue, Asp2267. The crystal structures of recombinant C2 domains from FV (Macedo-Ribeiro et al) and FVIII (Pratt et al) were superimposed, and residues around the mutated Asp2267 (FVIII) and the topologically equivalent Gln2132 (FV) are shown. The main chains of factors V and VIII are represented as orange and green ribbons, respectively. Only side chains of Gln2132/Asp2267 and surrounding residues are shown with all their nonhydrogen atoms; the side chains of FV are in orange and those of FVIII are color-coded as in Figure 3. Hydrogen bonds accepted from the Asp2267 carboxylate are indicated with dotted lines. Notice the complete equivalence of main-chain traces in the 2 cofactors, indicating that Asp2267 is dispensable for the observed loop conformation. Notice also that a large number of solvent-exposed side chains differ between the 2 cofactors, pointing to their involvement in specific protein-protein interactions. Inspection of the FVa/FVIIIa models suggests that these residues interact with substrates of the FVapi•FXa and FVIIIapi•FIXa complexes, prothrombin and FX, respectively.

Figure 5

Location of solvent-exposed mutations on the FVIIIa surface. Solid surface representation of the 2 hypothesized quaternary arrangements of human FVIIIa, the “compact” (A) and “extended” models (B). Domains are labeled, and the distances from the phospholipid membrane are indicated (calculated as minimum distances between the C-terminal Cys residue in domain A1, Cys329, to a plane passing through Cα atoms of membrane-binding residues in C2, Met2199, Phe2200, and Leu2252). The side chains of 4 novel mutations affecting exposed residues, Pro64, Ala375, Gly494, and Asp2267, are represented as light magenta spheres. In addition, several residues previously reported to participate in VWF binding are highlighted as yellow spheres; other reported but not characterized mutations affecting exposed residues are in red. A curved arrow points to the probable displacement of C-terminal residues from the a1 linker to a more extended conformation after cleavage of the Arg372-Ser373 peptide bond. This would bring further residues implicated in FX binding (eg, the triplet of acidic residues Asp361/Asp362/Asp363; Nogami et al) closer to the putative substrate binding site.