Crystallographic studies have previously suggested that Lys290 forms a salt bridge with Glu342 in the serine protease inhibitor alpha 1-antitrypsin. Disruption of the formation of this structural feature by a Glu to Lys substitution at residue 342 in the PiZ variant has been implicated in causing the defective secretion of this mutant protein from hepatocytes (10-15% of normal). To test the validity of this hypothesis, mutant human alpha 1-antitrypsin cDNA constructs coding for specific amino acid substitutions at residues 290 and 342 were generated and the corresponding mutant proteins were expressed in mouse hepatoma cells. When the potential to form the salt bridge was reestablished by a Lys290 to Glu290 substitution in the PiZ variant, its secretion was increased to only 38% of normal. Furthermore, disruption of this structural feature by a Lys290 to Glu290 substitution in the normal inhibitor failed to reduce the secretion of alpha 1-antitrypsin to the extent observed for the PiZ variant (73% of normal). Finally, substitution of the neutral amino acid Gln at residue 342 only reduced the secretion of alpha 1-antitrypsin to 55% of normal. Of all mutant proteins tested, those bearing Lys at position 342 were secreted at the lowest levels. These findings demonstrate that although disruption of the 290-342 salt bridge does affect the secretion of alpha 1-antitrypsin, it is the substitution of Lys at residue 342 that causes the dramatic secretory defect of the PiZ variant.