Factor (F)VIII functions as a cofactor in the Xase complex responsible for phospholipid (PL) surface-dependent conversion of FX to activated FX (FXa) by FIXa. Protein S (PS), a vitamin K-dependent anticoagulant protein, functions as a cofactor of activated protein C (APC) that inactivates FVIIIa and FVa. We have previously reported a new regulatory mechanism on coagulation that PS directly impaired the Xase complex by competing the FVIIIa-FIXa interaction (Takeyama, Br J Haematol. 2008), and identified at least two PS-interactive sites that were the 488-490 residues on FVIII A2 domain (Takeyama, Thromb Haemost. 2009) and the 2239 residue on FVIII C2 domain (Furukawa, ASH 2016). In the present study, to investigate the contribution of binding sites on FVIII A2 and C2 domains for the interaction between FVIII and PS, three recombinant B-domainless FVIII mutants substituted to alanine were expressed by BHK cell, such as the mutant of the 488-490 residues (S488A/R489A/R490A), the mutant of the 2239 residue (K2239A), and the mutant of both residues (S488A/R489A/R490A/K2239A), subsequently compared interaction with PS of the three mutants with wild-type (wt-) FVIII. First, we performed surface plasmon resonance-based assays to compare the direct binding of each FVIII mutant to PS. When PS was immobilized on the sensor chip and bound to the wt-FVIII and each FVIII mutant as an analyte, the K d value of wt-FVIII was 3.5 nM, while those of S488A/R489A/R490A, K2239A, and S488A/R489A/R490A/K2239A mutants were 43.7 nM, 6.7 nM, and 44.7 nM, respectively. These results demonstrated that the residues 488-490 and 2239 contributed to PS-interactive sites as described in the previous reports. Next, the effect of PS on the directly reduction of FVIIIa function was evaluated using the FXa generation assay with the three FVIII mutants and compared with wt-FVIII. After each FVIII mutant (1 nM) was reacted with PS (0-500 nM) in the presence of PL vesicles (20 µM) for 30 min, FVIII was activated by thrombin (30 nM), followed by the reaction with FIXa (40 nM) and FX (300 nM) for 1 min. Generated FXa in wt-FVIII was reduced in the presence of PS dose-dependently and was reduced to 7.7% at 125 nM of PS, compared to that in the absence of PS. However, generated FXa in K2239A, S488A/R489A/R490A, and S488A/R489A/R490A/K2239A were mildly reduced to 23.1 %, 23.6 %, and 29.2 % even at 500 nM of PS, respectively, compared to those in the absence of PS. Finally, to investigate the contribution of these FVIII and PS binding sites when PS plays as a cofactor of APC in FVIIIa inactivation, FXa generation assays were measured in each FVIII mutant in the presence of PS (150 nM) and APC (0-10 nM). In the presence of PS, generated FXa in wt-FVIII and each FVIII mutant were decreased in APC dose-dependently. At the 10 nM of APC, generated FXa in the wt-FVIII and K2239A were reduced to 48.1 % and 40.6 %, respectively, in comparison to the absence of APC, while generated FXa in S488A/R489A/R490A and S488A/R489A/R490A/K2239A mutants were mildly reduced to 65.8 % and 69.0 %, respectively. These findings suggested that the contribution of the 488-490 residues on the FVIII A2 domain was particularly great for the FVIII-PS interaction between the two PS binding sites identified on the FVIII A2 and C2 domains. In addition, the coexistence of the K2239A mutation on the FVIII C2 domain with S488A/R489A/R490A mutation further reduced the direct FVIII-PS interaction, suggesting that K2239A mutation further reduced the inhibitory effect of FVIII function by PS. On the other hand, mainly not the 2239 residue on the FVIII C2 domain but the 488-490 residues on the FVIII A2 domain were thought to affect the PS function as a cofactor in FVIIIa inactivation by APC. Overall, our data indicated that the 488-490 residues on the FVIII A2 domain were key residues for the FVIII-PS interaction including FVIIIa inactivation by APC, and the 2239 residue affected partially for only the FVIII-PS direct interaction.