We investigated the functional role of gamma-carboxyglutamic acid (Gla) residue 21 of human factor IX, using site-directed mutagenesis to change the glutamic acid residue to aspartic acid (FIX21D). FIX21D had reduced activity in an activated partial thromboplastin time (aPTT) assay and was activated by factor XIa more slowly than wild-type factor IX (FIXwt). FIX21D underwent normal, two-stage calcium-dependent intrinsic fluorescence quenching, indicating that a folding event similar to that seen in FIXwt occurred upon the addition of calcium ions. Antibody A-7, which recognizes factor IX-specific residues at positions 33-40, bound FIX21D as well as FIXwt; however, the calcium-specific monoclonal antibody, JK-IX-2, whose epitope includes residues 1 and 22, did not recognize FIX21D. FIX21D bound phosphatidylserine/phosphatidylcholine (PS/PC) vesicles with Kd approximately 10-fold greater than FIXwt, as measured by a fluorescence light scattering assay. Finally, although FIXwt binds endothelial cells with a Kd of 2.8 nM, FIX21D did not bind endothelial cells. Molecular modeling simulations of FIXwt and FIX21D indicate that mutating Gla 21 to Asp causes structural changes in residues 3-5 and 8-10, as well as in two exposed calcium ions, consistent with the reduced function of FIX21D. Immunological and intrinsic fluorescence quenching assays and the molecular dynamics simulations suggest normal folding in the C-terminal region of the Gla domain. Thus we hypothesize the FIX21D has reduced JK-IX-2 and phospholipid and endothelial cell binding due to localized structural changes in residues 3-10 and the exposed calcium ions. Our study suggests that the Gla 21 to Asp mutation disrupts function in the N-terminal region of the Gla domain without affecting structure in the C-terminal Gla domain region.