The ectonucleoside triphosphate diphosphohydrolases (NTPDases) control extracellularnucleotide concentrations, thereby modulating many important biological responses, including blood clottingand pain perception. NTPDases1-4 are oligomeric integral membrane proteins, whereas NTPDase5(CD39L4) and NTPDase6 (CD39L2) are soluble monomeric enzymes, making them more amenable tothorough structural and functional analyses than the membrane-bound forms. Therefore, we report herethe bacterial expression, refolding, purification, and biochemical characterization of the soluble portionof human NTPDase6. Consistent with the enzyme expressed in mammalian cells, this recombinantNTPDase6 efficiently hydrolyzes GDP, IDP, and UDP (specific activity of approximately 50000
molmg
-1 h
-1), with slower hydrolysis of CDP, ITP, GTP, CTP, ADP, and UTP and virtually no hydrolysisof ATP. The
Km for GDP (130 ± 30
M) is similar to that determined for the soluble rat NTPDase6expressed in mammalian cells. The secondary structure of the refolded enzyme was determined by circulardichroism to be 33%
-helix, 18%
-sheet, and 49% random coil, consistent with the secondary structurepredicted from the amino acid sequence of soluble NTPDase6. Four of the five cysteine residues in thesoluble NTPDase6 are highly conserved among all the NTPDases, while the fifth residue is not. Mutationof this nonconserved cysteine resulted in an enzyme very similar to wild type in its enzymology andsecondary structure, indicating that this cysteine exists as a free sulfhydryl and is not essential for structureor function. The disulfide pairing of the other four cysteine residues was determined as Cys
249-Cys
280and Cys
340-Cys
354 by HPLC and mass spectral analysis of tryptic peptides. Due to conservation of thesecysteine residues, these two disulfide bonds are likely to exist in all NTPDases. A structural model forNTPDase6, incorporating these and other findings obtained with other NTPDases, is proposed.