The binding to human serum proteins of MnDPDP (manganese(II) dipyridoxyl diphosphate), the active component of the magnetic resonance imaging contrast medium mangafodipir trisodium injection (Teslascan) was studied in ultrafiltration experiments. Sera from three males and three females were incubated with 86 microM [14C]MnDPDP for 60 min at room temperature (20-23 degrees C), followed by centrifugation through filters with a cut-off of 30 kDa. Analysis of the filtrates and the initial incubation mixtures for manganese, by ICP-AES, and for DPDP and its dephosphorylated metabolites DPMP (dipyridoxyl monophosphate) and PLED (dipyridoxyl ethylenediamine diacetate) by liquid scintillation counting, showed a clear difference in protein binding of manganese and the ligands under these conditions. Only 2.2 +/- 1.8% (mean +/- S.E.; n = 6) of DPDP, DPMP and PLED were bound to protein, whereas 26.9 +/- 2.9% (mean +/- S.E.; n = 6) of manganese was bound to protein. No binding of DPDP, DPMP or PLED to blood cells was observed when whole blood, containing either heparin or EDTA as anticoagulant, was spiked with [14C]MnDPDP and the cell-free fraction and the lysed cell fraction analysed by liquid scintillation counting. The extent of protein binding of manganese corresponded well with results from an in vitro metabolism study, in which MnDPDP was added to heparinized human whole blood, showing that approximately 25% of DPDP, DPMP or PLED were not bound to manganese. The in vitro metabolism study revealed that transmetallation with zinc was nearly complete within 1 min, and that dephosphorylation is a sequential process going from DPDP to the monophosphate DPMP, and then to the fully dephosphorylated compound PLED.