The use of a recombinant bacterial vector vaccine is an attractive vaccination strategy to induce an immune response to a carried protective antigen. The superiorities of live bacterial vectors include mimicry of a natural infection, intrinsic adjuvant properties, and the potential for administration by mucosal routes. Escherichia coli is a simple and efficient vector system for production of exogenous proteins. In addition, many strains are nonpathogenic and avirulent, making it a good candidate for use in recombinant vaccine design. In this study, we screened 23 different iron-regulated promoters in an E. coli BL21(DE3) vector and found one, P(viuB), with characteristics suitable for our use. We fused P(viuB) with lysis gene E, establishing an in vivo inducible lysis circuit. The resulting in vivo lysis circuit was introduced into a strain also carrying an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible P(T7)-controlled protein synthesis circuit, forming a novel E. coli-based protein delivery system. The recombinant E. coli produced a large amount of antigen in vitro and could deliver the antigen into zebrafish after vaccination via injection. The strain subsequently lysed in response to the iron-limiting signal in vivo, implementing antigen release and biological containment. The gapA gene, encoding the protective antigen GAPDH (glyceraldehyde-3-phosphate dehydrogenase) from the fish pathogen Aeromonas hydrophila LSA34, was introduced into the E. coli-based protein delivery system, and the resultant recombinant vector vaccine was evaluated in turbot (Scophtalmus maximus). Over 80% of the vaccinated fish survived challenge with A. hydrophila LSA34, suggesting that the E. coli-based antigen delivery system has great potential in bacterial vector vaccine applications.