A cDNA encoding soybean alpha-D-galactosidase [E.C. 3.2.1.22] was obtained by screening a soybean library with Phaseolus alpha-D-galactosidase cDNA. The Glycine max alpha-D-galactosidase cDNA is 1.75 kb long and contains untranslated 5' and 3' sequences. The deduced amino acid sequence of the soybean gene has a high degree of homology with other eucaryotic alpha-D-galactosidases. Recombinant alpha-D-galactosidase (rGal) was expressed in Pichia pastoris and purified by affinity chromatography. Purified rGal was homogeneous as judged by SDS-PAGE analysis with the relative molecular mass under reducing conditions of 39.8, and under nonreducing conditions 38.0 kDa. The expressed protein contained the sequence NGLGHTPPMG at the N-terminus, corresponding to the deduced amino acid sequence of the soybean gene. The relative native molecular mass by Sephacryl S-200 chromatography was determined to be 33.1 kDa. The specific activity was 295.6 mumoles of PNP-alpha-D-galactopyranoside hydrolyzed per mg pure rGal per min. rGal was highly specific for alpha-D-galactosyl residues. No detectable hemagglutinin or protease activity was present in the preparations. Furthermore, rGal was active against the blood group B antigen in native human erythrocyte cell suspension assays. The only detectable erythrocyte phenotypic change was loss of the B and P1 epitopes. Consequently, recombinant Glycine max alpha-D-galactosidase may have useful biotechnical applications in the potential mass production of universally transfusable type O erythrocytes by enzymatic conversion.