Bacterial Outer Membrane Vesicles (OMVs) contribute to virulence, competition, immune avoidance and communication. This has led to great interest in how they are formed. To date, investigation has focused almost exclusively on what controls the initiation of OMV biogenesis. Regardless of the mechanism of initiation, all species face a similar challenge before an OMV can be released: How does the OM detach from the underlying peptidoglycan (PG) in regions that will ultimately bulge and then vesiculate? The OmpA family of OM proteins (OprF in P. aeruginosa) is widely conserved and unusually abundant in OMVs across species considering their major role in PG attachment. OmpA homologs also have the interesting ability to adopt both PG-bound (two-domain) and PG-released (one-domain) conformations. Using targeted deletion of the PG-binding domain we showed that loss of cell wall association, and not general membrane destabilization, is responsible for hypervesiculation in OprF-modified strains. We therefore propose that OprF functions as a 'latch', capable of releasing PG in regions destined to become OMVs. To test this hypothesis, we developed a protocol to assess OprF conformation in live cells and purified OMVs. While >90% of OprF proteins exist in the two-domain conformation in the OM of cells, we show that the majority of OprF in OMVs is present in the one-domain conformation. With this work, we take some of the first steps in characterizing late-stage OMV biogenesis and identify a family of proteins whose critical role can be explained by their unique ability to fold into two distinct conformations.