Plastids with four envelope membranes have evolved from red and green algae engulfed by phagotrophic protozoans. It is assumed that the Sec translocon resides in their outermost membrane, while in the two innermost membranes the Toc-Tic supercomplex is embedded. However, such a single Sec/single Toc-Tic model cannot explain the passage of proteins across the second (or periplastid) membrane which represents the endosymbiont plasmalemma. One of the most recent models postulates that this membrane contains the Toc75 channel which was relocated here from the endosymbiont plastid. Unfortunately, the precursor of this protein carries a bipartite presequence, which means that its insertion into the new membrane would require relocation and/or modification of two different processing peptidases. I suggest that these obstacles can be easily bypassed by the assumption that the mitochondrial Tim23 channel was inserted into the endosymbiont plasmalemma. In contrast to Toc75, this protein has an internal, uncleavable targeting signal and its insertion into the new membrane would require neither relocation nor modification of additional proteins. Besides, such a relocated Tim23 channel could import not only plastid, but also mitochondrial proteins. I hypothesize that from the latter proteins, initially directed to the endosymbiont mitochondrion, periplastid proteins have evolved which are now targeted to the former cytosol and/or nucleus of the eukaryotic algal endosymbiont.