The formation of protein disulfide bonds in the Escherichia coli periplasm by the enzyme DsbA is an inaccurate process. Many eukaryotic proteins with nonconsecutive disulfide bonds expressed in E. coli require an additional protein for proper folding, the disulfide bond isomerase DsbC. Here we report studies on a native E. coli periplasmic acid phosphatase, phytase (AppA), which contains three consecutive and one nonconsecutive disulfide bonds. We show that AppA requires DsbC for its folding. However, the activity of an AppA mutant lacking its nonconsecutive disulfide bond is DsbC-independent. An AppA homolog, Agp, a periplasmic acid phosphatase with similar structure, lacks the nonconsecutive disulfide bond but has the three consecutive disulfide bonds found in AppA. The consecutively disulfide-bonded Agp is not dependent on DsbC but is rendered dependent by engineering into it the conserved nonconsecutive disulfide bond of AppA. Taken together, these results provide support for the proposal that proteins with nonconsecutive disulfide bonds require DsbC for full activity and that disulfide bonds are formed predominantly during translocation across the cytoplasmic membrane.Disulfide bonds between cysteine residues make an important contribution to the folding pathway and stability of many proteins. Early in vitro studies on protein folding showed that the protein RNase when denatured and its disulfide bonds reduced could refold into an active enzyme with the correct disulfide bonds formed (1). Although these results indicated that the information necessary for the proper folding of proteins was intrinsic to the amino acid sequence, the rate of disulfide bond formation in these experiments was much slower than the rates observed in living cells, and the yield of active enzyme was low. The low yields of properly folded enzyme were attributed to the formation of incorrect disulfide bonds in many of the assembled protein molecules. Thus, the in vitro studies, although remarkably successful, were unable to come close to replicating the rapid kinetics and the accuracy of in vivo disulfide bond formation. To explain this inaccuracy observed in biochemical experiments, Anfinsen and co-workers